02 / Models and capability index
AI Model Directory
Showing 317 of 317 records
Distinguish named models, families and product capability groups, then find official sources by vendor, modality and task.
Reviewed in this update
Search the model directory ↓An editorial selection across vendors and modalities, not a capability ranking. Dates describe source checks, not model releases. Every record remains searchable below.
OpenAI
GPT-6.1 Sol
The newer Sol model for complex coding, computer use, and professional work. Tool calling uses Responses; Chat Completions supports requests without tool calls.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Opus 5.5
The current Opus for long-running agentic coding and knowledge work, released 2026-09-22. It accepts text/images, returns text, and uses always-on adaptive thinking.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 3.8 Flash
A stable Flash for long-horizon software engineering, agents, and enterprise workflows. It accepts text, images, video, audio, and PDFs, and outputs text, not audio or images.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
DeepSeek
DeepSeek V4.1 Flash
The current deepseek-flash API serves V4.1 Flash for image understanding, text reasoning, coding and tool calls.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Alibaba
Qwen3.8-Max
A vision-language model for software engineering, professional documents and long-horizon agents, understanding images and video with text output.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Moonshot AI
Kimi K3
A reasoning model for visual understanding, coding and knowledge work, accepting images and video; thinking stays enabled with adjustable effort.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Black Forest Labs
FLUX 3 Image
Image generation and editing with bounding-box layout control, accessible through Playground or BFL API. This does not establish released Dev weights.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ByteDance
Seedance 2.5
A joint audio-video generation model using reference-video framing and cinematic intent, with audiovisual editing, green-screen editing and video extension.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ElevenLabs
Eleven v4
Expressive speech and multi-speaker dialogue model with 90+ languages and voice cloning, documented for the Text to Dialogue API.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Meshy
Meshy 7.1
Added September 18, 2026, for text, image and multi-image 3D APIs. The documented latest alias resolves to meshy-7.1.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Haiku 5.5
Released 2026-10-07 for high-volume, latency-sensitive classification, extraction, routing, and subagent tasks, with text/image input, tools, and configurable adaptive thinking.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Mistral AI
Mistral Large 4
An open-weight multimodal mixture-of-experts model supporting document Q&A, function calling and structured outputs. Check its version-specific license.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
All directory records
Search names, aliases, vendors or tasks, then narrow the results. Legacy tool-link counts do not determine the order.
Showing 317 of 317 records
Adobe
Firefly Image Model 5
Adobe's image generation and editing model. The March 19, 2026 announcement confirms general availability beyond its earlier public beta.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Alibaba
Qwen-Image 3.0 Pro
Generates and edits images, with an official focus on complex layouts, small lettering and multilingual text rendering.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Alibaba
Qwen3.7-Plus
A general model for codebases, office documents and visual understanding, combining reasoning, tool calls and structured output for multi-step tasks.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Alibaba
Qwen3.8-Flash
The Flash model in Qwen3.8, for everyday documents, coding and visual questions, with tool calling and structured text output.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Alibaba
Qwen3.8-Max
A vision-language model for software engineering, professional documents and long-horizon agents, understanding images and video with text output.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Alibaba
Qwen3.8-Omni-Flash
Analyzes text, images, audio and video and returns text; speech output requires a separate realtime or speech model.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Alibaba
Qwen3.8-Omni-Flash-Realtime
For realtime interaction with text, streaming audio and video-frame input, returning text and speech with tool-call support.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Alibaba
Wan2.7-Image-Pro
An image generation and editing model with palette control, multi-image references, bounding-box edits and consistent-character images.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Alibaba
Wan3.0-Video
Unifies text-to-video, frame control and multimodal references, generating dialogue, music and effects while supporting video editing and extension.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Fable 5.1
For demanding reasoning, long-horizon agents, research, and document, spreadsheet, and slide work. Released 2026-09-01; distinct from verification-gated Mythos 5.1.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Haiku 5.5
Released 2026-10-07 for high-volume, latency-sensitive classification, extraction, routing, and subagent tasks, with text/image input, tools, and configurable adaptive thinking.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Opus 5.5
The current Opus for long-running agentic coding and knowledge work, released 2026-09-22. It accepts text/images, returns text, and uses always-on adaptive thinking.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Sonnet 5.5
The current Sonnet, released 2026-09-28, balances speed and capability with vision, tool use, and adaptive thinking. Existing Sonnet 5 integrations need a migration review.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Black Forest Labs
FLUX 3 Image
Image generation and editing with bounding-box layout control, accessible through Playground or BFL API. This does not establish released Dev weights.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Black Forest Labs
FLUX 3 Video
Video generation with native audio, multiple shots and keyframes, available through dashboard and API. Action access and Dev weights have separate rollouts.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ByteDance
Doubao Seed Evolving
Ark's continuously updated coding and agent model service, with visual understanding and tool calls; a rolling model ID rather than a fixed snapshot.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ByteDance
Doubao-Seed-2.1-Lite
The 260915 Lite model, officially listed for visual and audio understanding, reasoning, tool calls and structured text output.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ByteDance
Doubao-Seed-2.1-Pro
The 260915 Pro model recommended by Ark, with multimodal understanding, deep reasoning, GUI tasks and tool calls.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ByteDance
Doubao-Seed-2.1-Turbo
Ark's 260628 Turbo model for multimodal understanding, GUI and tool tasks, also offered through a low-latency inference service.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ByteDance
Seed Audio 1.0
Generates dialogue, effects and ambience from scene prompts, with voice design from text or authorized samples and control over dialogue timing.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ByteDance
Seedance 2.5
A joint audio-video generation model using reference-video framing and cinematic intent, with audiovisual editing, green-screen editing and video extension.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ByteDance
SeedRealtime
A native full-duplex audio-visual model that follows continuous streams, times spoken responses and supports proactive reminders and tool interactions.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ByteDance
Seedream 5.0 Pro
For text-rich infographics and complex image creation, with spatial annotations, sketch-guided interactive editing and layer separation.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
DeepSeek
DeepSeek V4 Pro
A text reasoning, coding and tool-calling model. The current official API version is V4-Pro-0813 and does not accept vision input.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
DeepSeek
DeepSeek V4.1 Flash
The current deepseek-flash API serves V4.1 Flash for image understanding, text reasoning, coding and tool calls.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ElevenLabs
Eleven Music v2.5
Text-to-music with composition plans, audio references and inpainting. Its API ID is music_v2_5; v2 remains available.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ElevenLabs
Eleven v4
Expressive speech and multi-speaker dialogue model with 90+ languages and voice cloning, documented for the Text to Dialogue API.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 3.1 Pro Preview
A multimodal reasoning Pro preview taking text, images, video, audio, and PDFs and returning text, with code execution and tools. Its API ID retains preview.
Official identity and purpose checked
Preview / testing access
Sources checked:
Gemini 3.5 Flash-Lite
A stable lightweight model for high-throughput work, understanding text, images, video, and audio and producing text for large volumes of defined tasks.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 3.5 Transcribe
Transcribes audio files with language detection, speaker diarization, and word timestamps. Custom vocabulary cannot be combined with diarization or timestamps.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 3.8 Flash
A stable Flash for long-horizon software engineering, agents, and enterprise workflows. It accepts text, images, video, audio, and PDFs, and outputs text, not audio or images.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 3.8 Flash TTS
Turns scripts into speech for audiobooks, narration, and multi-speaker dialogue. Specify delivery and speakers in speech_metadata rather than copying older inline directions.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 3.8 Flash-Lite TTS
A lightweight text-to-speech model for bulk production, read-aloud, and real-time voice-agent cascades. It shares Flash TTS's structured request format, with a separate model ID.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 3.8 Live
A stable model for low-latency live voice dialogue, with audio streaming, asynchronous tools, and session updates, accepting voice and visual context and responding in audio.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 3.8 Live Extended Thinking
A stable Live model that reasons in the background and calls asynchronous tools while streaming audio, for complex multi-step voice tasks. Session handling differs from standard 3.8 Live.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini Omni Flash
The stable gemini-omni-1.1-flash for conversational video creation and editing from text, images, or existing video, including extension, interpolation, and upscaling. It is not a live-chat model.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Nano Banana 2.1
A stable image generation/editing model updated in October 2026, using gemini-nano-banana-2.1, with a focus on visual quality, typography, and multi-turn consistency.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Nano Banana Pro
An image generation/editing model with API ID gemini-3-pro-image, accepting text/images and returning images/text. It is distinct from Nano Banana 2 and the original Nano Banana.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Google DeepMind
Lyria 3.5
A stable model generating full songs from text prompts or images, with verses, choruses, and bridges, returning music and lyrics. It is separate from realtime music streaming.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Google DeepMind
Veo 3.1
Generates video with synchronized audio from text or images for controlled cinematic storytelling. This record covers the Gemini API veo-3.1-generate-preview endpoint.
Official identity and purpose checked
Preview / testing access
Sources checked:
Google DeepMind
Veo 3.1 Lite Preview
A Lite preview for high-volume video applications, generating video with audio from text or images. Unlike other Veo 3.1 tiers, it supports neither 4K nor video extension.
Official identity and purpose checked
Preview / testing access
Sources checked:
Ideogram
Ideogram 4.5
The vendor's current image editing model for prompt-guided changes that preserve the source image, separate from Ideogram 4's open-weight release.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Lightricks
LTX-2.3
Audio-video generation weights with local inference, Dev and distilled variants and upsamplers, subject to the LTX-2 Community License.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Luma
Ray 3.2
Video generation, editing and reframing on Luma Agents API. Legacy Ray retirement cutoffs depend on each customer's notice.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Meshy
Meshy 7.1
Added September 18, 2026, for text, image and multi-image 3D APIs. The documented latest alias resolves to meshy-7.1.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Meta
Muse Spark 1.3
Released September 2, 2026, for longer agentic and coding tasks, available in Muse Code and Meta Model API. Open weights remain a future plan.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Midjourney
Midjourney V8.2
The version guide lists V8.2 as the current default, introduced July 24, 2026, with the Edit Model and updated personalization.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
MiniMax
MiniMax M3
A native vision-language model for long-horizon coding, chart understanding and agent tool use, offered through the MiniMax-M3 API.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
MiniMax
MiniMax Speech 2.8
A speech-synthesis model with natural delivery tags, voice cloning and cross-language synthesis for expressive spoken narration.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Mistral AI
Mistral Large 4
An open-weight multimodal mixture-of-experts model supporting document Q&A, function calling and structured outputs. Check its version-specific license.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Moonshot AI
Kimi K2.6
Accepts text, images and video for conversation, coding and agent tasks, with both thinking and non-thinking modes.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Moonshot AI
Kimi K2.7 Code
A thinking model for long-horizon coding, visual inputs and multi-step tool calls; HighSpeed serves the same model at higher speed.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Moonshot AI
Kimi K3
A reasoning model for visual understanding, coding and knowledge work, accepting images and video; thinking stays enabled with adjustable effort.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-6 Astra
A flagship for complex reasoning, coding, and tool use. It takes text and images and returns text; its API ID is gpt-6-astra.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-6 Luna
A lightweight GPT-6 model for focused, high-volume workloads. It takes text and images and returns text; use gpt-6-luna.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-6.1 Sol
The newer Sol model for complex coding, computer use, and professional work. Tool calling uses Responses; Chat Completions supports requests without tool calls.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-Image-2.5 Flare
GPT Image 2.5 for quick everyday image generation and editing from text and images. Flare and Sunburst use different API IDs.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-Image-2.5 Sunburst
The capability-focused GPT Image 2.5 model for generation and editing, taking text and reference images and returning images. A 2026-09-08 snapshot is documented.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-Live 1
A full-duplex model for live voice conversations, supporting simultaneous listening and speaking, interruptions, and delegation of reasoning and tools to a backend agent.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-Live-Transcribe
Streams low-latency transcript deltas from live audio for captions and transcription interfaces, with tunable latency and context, keyword, and language hints.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-Realtime-2.1
A speech-to-speech model for complex voice agents, with configurable reasoning and tools plus improved alphanumeric recognition, silence, noise, and interruption handling.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-Realtime-2.1 Mini
A distilled reasoning model for faster, lower-cost real-time voice interactions, accepting audio and text over WebRTC, WebSocket, or SIP.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-Transcribe
Transcribes completed audio files, streamed files, and committed realtime turns, using context, keyword, and language hints. It does not synthesize speech.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Pika
Pika 2.5
Generates video from text, an image or ordered keyframes. Output is silent; Pika Soundtrack is a separate audio step.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Recraft
Recraft V4.1
Image and native-vector generation family with standard, Vector, Pro and Utility variants. Choose raster or SVG output explicitly.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Recraft
Recraft V4.1 Flash
A fast image model announced September 23, 2026, for visual iteration, with an enhancement workflow using V4.1 Pro.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Runway
Runway Aleph 2.0
In-context video editing: edit a frame and propagate changes across relevant shots. The vendor documents clips up to 30 seconds at 1080p.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
xAI
Grok 4.7
The current flagship for coding, agents, and knowledge work, accepting text/images and returning text, with tools, structured outputs, and configurable reasoning.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
xAI
Grok Imagine Image 2.0
Generates images from text or edits reference images with configurable quality and resolution. Use grok-imagine-image-2.0, distinct from the older Image Quality model.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
xAI
Grok Imagine Video 1.5
Generates video from text or images, with audio by default and lip-synced speech, plus visual and voice references. Video jobs are asynchronous and require waiting for completion.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
xAI
Grok Imagine Video 1.5 Lite
A separate lightweight Video 1.5 model generating lip-synced video from text or a single image. Use grok-imagine-video-1.5-lite; do not assume all standard-model reference inputs.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Z.ai
GLM-5.3
A text model for software engineering and long-horizon agents, supporting tool calls and structured output with reasoning always enabled.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Z.ai
GLM-5.3-Flash
A visual coding model combining image, video and file understanding with tools for interface development and document work; outputs text.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Z.ai
GLM-5.3-FlashX
A faster multimodal GLM-5.3 API model for visual understanding and tool tasks; official documentation excludes it from GLM Coding Plan currently.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Haiku 4.5
An Active (legacy) Haiku whose alias resolves to claude-haiku-4-5-20251001. It supports manual extended thinking, distinct from Haiku 5.5's adaptive thinking.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Mythos 5.1
Shares Fable 5.1's capabilities and specifications, but access is limited to organizations approved through Anthropic verification programs. Active does not mean access for every account.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Opus 4.6
An older Opus still served with text/image input and text output. Anthropic lists it as legacy, rather than the current recommended flagship.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Opus 4.8
A still-served legacy Opus with text/image input and text output for existing workflows. New model choices should also consider the current Opus 5.5.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Opus 5
A previous Opus still marked Active (legacy), with text/image input and text output. Opus 5.5 is current; legacy does not mean shut down.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Sonnet 4.6
A still-available legacy Sonnet with text/image input and text output, separate from deprecated Sonnet 4.5 and current Sonnet 5.5.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Anthropic
Claude Sonnet 5
A previous Sonnet still served as a legacy model. It is not an alias for Sonnet 5.5, and the newer release does not itself retire this API.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ElevenLabs
Eleven Flash v2.5
Low-latency speech synthesis. The vendor recommends Flash over corresponding Turbo models; their model identifiers remain distinct.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ElevenLabs
Eleven v3
A previous-generation expressive voice model still listed by the vendor, with 70+ languages. Previous generation does not mean retired.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ElevenLabs
Eleven v4 Turbo
The v4 real-time speech model uses the Text to Dialogue WebSocket, a different interface from standard v4.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
ElevenLabs
Scribe v2
Speech recognition in 90+ languages with word timestamps and speaker diarization. This is transcription, not speech synthesis.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 2.5 Flash
An older stable Flash still serving text, image, video, and audio understanding with text output. Not deprecated, but access is currently restricted to previous active 2.5 users.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 2.5 Pro
An older stable Pro for complex reasoning, coding, and long-document analysis. Not deprecated, but Google currently limits 2.5 access to previous active users.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 3.5 Transcribe Live
Transcribes live audio over WebSocket with language detection and custom vocabulary. Unlike the file endpoint, Live supports neither diarization nor word timestamps.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 3.6 Flash
A still-served previous stable Flash for everyday multimodal and agentic tasks, understanding text, images, video, and audio and returning text.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini 3.7 Flash
A confirmed older Flash. It is no longer selectable in AI Studio, but the gemini-3.7-flash endpoint remains open and automatically routes to gemini-3.8-flash.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Gemini Embedding 2
Maps text, images, video, audio, and PDFs into one embedding space for cross-modal search, document retrieval, and recommendations. The stable gemini-embedding-2 is not a chat-generation model.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Nano Banana
The original Nano Banana uses gemini-2.5-flash-image for image generation and editing. Google's deprecation table lists an earliest shutdown date of 2027-03-15.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Nano Banana 2
Uses gemini-3.1-flash-image for fast, high-volume image generation and conversational editing, with search grounding. It is a different model from Pro.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Nano Banana 2 Lite
A lightweight model for interactive images and multi-turn local edits such as colors, stickers, and backgrounds. The documented image size is 1K, without 2K or 4K support.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Google DeepMind
Lyria 3 Clip Preview
Generates 30-second musical clips, loops, and previews from text or images, returning audio and lyrics. This is lyria-3-clip-preview, separate from Lyria 3 Pro.
Official identity and purpose checked
Preview / testing access
Sources checked:
Google DeepMind
Veo 3.1 Fast
A separate Veo 3.1 Fast preview endpoint, veo-3.1-fast-generate-preview, generating video with audio from text or images. Its ID is separate from the standard model.
Official identity and purpose checked
Preview / testing access
Sources checked:
Ideogram
Ideogram 4
Text-to-image model released June 3, 2026, with JSON layout and palette controls. Downloadable weights are gated and use a non-commercial license.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Meshy
Meshy 6 Lite
The API migration target for Meshy 5, retired October 10, 2026. New requests use meshy-6-lite; existing generated assets remain available.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Midjourney
Midjourney V8.1
Released April 14, 2026, with native 2K HD images. It was a previous default; V8.2 is now the default version.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Midjourney / Spellbrush
Niji 7
An anime and illustration model launched January 9, 2026, developed with Spellbrush, with updated line work, detail and prompt following.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
MiniMax
MiniMax M2.7
An agent model for software engineering, log debugging and Office document editing; standard and highspeed APIs serve the same model at different speeds.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Mistral AI
Mistral Medium 3.5
A multimodal model for agentic and coding tasks, version 26.04, under Modified MIT rather than Small 4's license.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Mistral AI
Mistral OCR 4.1
Document OCR with paragraph bounding boxes, structural block labels and block confidence scores, intended for document extraction.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Mistral AI
Mistral Small 4
A hybrid model unifying instruction following, reasoning and coding. Version 26.03 is Apache 2.0, distinct from the generic Mistral Small family label.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-5.6 Luna
GPT-5.6 for cost-sensitive, high-volume tasks, with text/image input, text output, and configurable reasoning. It is separate from GPT-6 Luna.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-5.6 Sol
The Sol tier of GPT-5.6 for complex professional work. The gpt-5.6 alias routes here; Terra and Luna have separate model IDs.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-5.6 Terra
A separate GPT-5.6 tier balancing capability and cost, with text/image input, text output, and configurable reasoning effort. It is not a Sol alias.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
OpenAI
GPT-6 Sol
GPT-6 Sol supports complex coding and agentic workflows. It is separate from the newer GPT-6.1 Sol, with a distinct API ID.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Pika
Pika Soundtrack
A video-to-audio model announced August 18, 2026, generating synchronized effects, ambience, voice and music for uploaded video.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Recraft
Recraft V4 Styles
A family that builds reusable styles from 1–10 references, with raster, Vector and Pro variants for consistent visual assets.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
xAI
Grok 4.3
An earlier model with text/image understanding, tools, and configurable reasoning. Some retired Grok text IDs route here; a resolving old ID does not mean its original model still runs.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
xAI
Grok 4.6
Released 2026-08-12 for long-running agents, coding, and interactive visual projects. A separate grok-4.6 API model is documented; Grok 4.7 is the current recommended flagship.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
Z.ai
GLM-5.2
A text model for long-horizon software engineering and agents, with tools and structured output; distinct from visual Flash models.
Official identity and purpose checked
Listed as available by vendor
Sources checked:
MiniMax
MiniMax H3
Understands text, image, video and sound references for joint audio-video generation, motion transfer and multimodal content editing.
Official identity and purpose checked
Availability not confirmed
Sources checked:
MiniMax
MiniMax Music 3.0
Creates complete songs from musical ideas and optional lyrics, combining composition, arrangement, singing and audio production in one generation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
AIVA
AIVA model
AIVA model is a directory label for AIVA's music and song generation product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Alibaba
Qwen3
Qwen3 is an Alibaba model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Alibaba
Qwen3.8
A family including Max and Flash for coding, documents and visual understanding; choose capabilities by the full model ID.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Anthropic
Claude 3
Claude 3 is an Anthropic model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Anthropic
Claude 3.5
Claude 3.5 is an Anthropic model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Anthropic
Claude family
Claude family is an Anthropic model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Anthropic
Claude Opus 4.5
Claude Opus 4.5 is an Anthropic model for reasoning, coding, and tool use. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Anthropic
Claude Opus 4.7
Claude Opus 4.7 is an Anthropic model for reasoning, coding, and tool use. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Augment Code
Augment model
Augment model is a directory label for Augment Code's software development assistance product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Baidu
ERNIE 4.x
ERNIE 4.x is a Baidu model-family or version-line entry for text understanding and generation; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Beatoven.ai
Beatoven model
Beatoven model is a directory label for Beatoven.ai's music and song generation product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Beautiful.ai
Beautiful AI
Beautiful AI is a directory label for Beautiful.ai's presentation generation and layout product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Black Forest Labs
FLUX family
FLUX family is a Black Forest Labs model-family or version-line entry for image generation and editing; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Black Forest Labs
FLUX.2
FLUX.2 is a Black Forest Labs model-family or version-line entry for image generation and editing; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Black Forest Labs
FLUX.2 Klein
FLUX.2 Klein is a Black Forest Labs model-family or version-line entry for image generation and editing; exact snapshots, API availability, and licensing must be checked in the official documentation. The official page distinguishes 4B/9B and Base/distilled variants. Their licenses differ; the whole family is not unrestricted commercial open source.
Official identity and purpose checked
Availability not confirmed
Sources checked:
ByteDance
CapCut AI models
CapCut AI models is a directory label for ByteDance's video editing product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
ByteDance
Doubao models
Doubao models served by Volcano Engine Ark, including Seed2.1 Pro, Lite and Turbo; audio understanding and other capabilities vary by model.
Official identity and purpose checked
Availability not confirmed
Sources checked:
ByteDance
Seed2.1
The Pro and Turbo family described by ByteDance Seed for office agents, software engineering, image and video understanding.
Official identity and purpose checked
Availability not confirmed
Sources checked:
ByteDance
Seedance
ByteDance Seed's video family. Version 2.5 supports joint audio-video generation, references, editing and extension; capabilities depend on the version.
Official identity and purpose checked
Availability not confirmed
Sources checked:
ByteDance
Seedream
ByteDance Seed's image generation and editing family; 4.0 and 5.0 Pro are separate versions whose features should be compared by full name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
ClickUp
ClickUp Brain
ClickUp Brain is a directory label for ClickUp's project and document assistance product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Cognition
Devin model
Devin model is a directory label for Cognition's software development assistance product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
D-ID
D-ID avatars
D-ID avatars is a directory label for D-ID's avatar video and voiceovers product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
DeepL
DeepL translation model
DeepL translation model is a directory label for DeepL's translation product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
DeepSeek
DeepSeek family
DeepSeek family is a DeepSeek model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Descript
Descript AI
Descript AI is a directory label for Descript's video editing product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
ElevenLabs
ElevenLabs TTS
ElevenLabs TTS is an ElevenLabs model-family or version-line entry for speech synthesis; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Exa
Exa embedding model
Exa embedding model is a directory label for Exa's web retrieval product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Fathom
Fathom model
Fathom model is a directory label for Fathom's meeting transcription and summaries product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Fireflies.ai
Fireflies model
Fireflies model is a directory label for Fireflies.ai's meeting transcription and summaries product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Fish Audio
Fish Speech
Fish Speech is a Fish Audio model-family or version-line entry for speech synthesis; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Fish Audio
OpenAudio S2
OpenAudio S2 is a Fish Audio model for speech synthesis. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Flair AI
Flair model
Flair model is a directory label for Flair AI's product-image creation and editing product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Fliki
Fliki Voice
Fliki Voice is a directory label for Fliki's speech synthesis product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Gamma
Gamma AI
Gamma AI is a directory label for Gamma's presentation generation and layout product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Genmo
Mochi 1
Mochi 1 is a Genmo model for video generation. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Glean
Glean model
Glean model is a directory label for Glean's enterprise search and question answering product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Gemini family
Gemini family is a Google model-family or version-line entry for text and image understanding; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Google DeepMind
Lyria 3 family
A music-generation family entry, not one API ID. The official catalog distinguishes Lyria 3 Clip, the previous Lyria 3 Pro, and the separate Lyria 3.5.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Grammarly
Grammarly AI
Grammarly AI is a directory label for Grammarly's writing assistance product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
HeyGen
HeyGen AI
HeyGen AI is a directory label for HeyGen's avatar video and voiceovers product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Ideogram
Ideogram 1.0
Ideogram 1.0 is an Ideogram model for image generation and editing. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Ideogram
Ideogram 2.0
Ideogram 2.0 is an Ideogram model for image generation and editing. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Ideogram
Ideogram 3
Ideogram 3 is an Ideogram model for image generation and editing. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Inflection
Pi model
Pi model is a directory label for Inflection's conversational interaction product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Interior AI
Interior AI model
Interior AI model is a directory label for Interior AI's interior-design visualization product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
InVideo
InVideo model
InVideo model is a directory label for InVideo's video editing product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Kits.AI
Kits Voice
Kits Voice is a directory label for Kits.AI's voice conversion and voice-model training product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Krea
Krea Realtime
Krea Realtime is a directory label for Krea's image generation and editing product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Krisp
Krisp model
Krisp model is a directory label for Krisp's call noise cancellation and meeting notes product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Leonardo AI
Leonardo Phoenix
Leonardo Phoenix is a Leonardo AI model for image generation and editing. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
LTX Studio
LTX Video
LTX Video is a LTX Studio model-family or version-line entry for video generation; exact snapshots, API availability, and licensing must be checked in the official documentation. The official repository documents text-to-video and image-to-video and distinguishes weight versions and distilled variants.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Magnific AI
Magnific Upscaler
Magnific Upscaler is a directory label for Magnific AI's image upscaling and restoration product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Mem
Mem model
Mem model is a directory label for Mem's note organization and retrieval product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Meta
Llama 4
Llama 4 is a Meta model-family or version-line entry for text and image understanding; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Meta
Llama family
Llama family is a Meta model-family or version-line entry for text and image understanding; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Microsoft
Copilot model
Copilot model is a directory label for Microsoft's project and document assistance product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Midjourney
Midjourney V7
Midjourney V7 is a Midjourney model-family or version-line entry for image generation and editing; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Midjourney
Midjourney V8
The V8 family includes V8.1 and V8.2. Early V8.0 Alpha became unavailable on July 24, 2026; saved images remain. The whole V8 family is not retired.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Mistral
Codestral
Codestral is a Mistral model-family or version-line entry for software development assistance; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Mistral
Mistral Large
Mistral Large is a Mistral model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Mistral
Mistral Large 3
Mistral Large 3 is a Mistral model for text and image understanding. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Mistral
Mistral Small
Mistral Small is a Mistral model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Mistral
Voxtral
Voxtral is a Mistral model-family or version-line entry for audio recognition and understanding; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Moises
Moises model
Moises model is a directory label for Moises's vocal and instrument separation product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Moonshot AI
Kimi model
Moonshot AI's model family. Current APIs include K3, K2.7 Code and K2.6 for reasoning, coding, image and video understanding.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Motion
Motion scheduler
Motion scheduler is a directory label for Motion's calendar and task scheduling product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Move AI
Move model
Move model is a directory label for Move AI's markerless motion capture product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Mubert
Mubert model
Mubert model is a directory label for Mubert's music and song generation product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Murf
Murf Voice
Murf Voice is a directory label for Murf's speech synthesis product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Napkin AI
Napkin model
Napkin model is a directory label for Napkin AI's text-to-diagram creation product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Nous Research
Hermes
Hermes is a Nous Research model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Open-source community
Real-ESRGAN
Real-ESRGAN is an Open-source community model for image upscaling and restoration. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
OpenAI
DALL·E family
DALL·E family is an OpenAI model-family or version-line entry for image generation and editing; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
OpenAI
GPT family
GPT family is an OpenAI model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
OpenAI
GPT Image 2.5 family
An image generation and editing family with separate Sunburst and Flare models, accepting text or reference images. The older gpt-image-1 is not its alias.
Official identity and purpose checked
Availability not confirmed
Sources checked:
OpenAI
GPT-4
GPT-4 is an OpenAI model for text understanding and generation. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. This entry refers to the documented legacy text GPT-4 model, not GPT-4o or another multimodal variant.
Official identity and purpose checked
Availability not confirmed
Sources checked:
OpenAI
gpt-oss
gpt-oss is an OpenAI model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation. The consulted official card is gpt-oss-120b; check open-weight deployment and hosted API support separately.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Opus Clip
Opus model
Opus model is a directory label for Opus Clip's long-video clipping and short-video editing product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Otter.ai
Otter model
Otter model is a directory label for Otter.ai's meeting transcription and summaries product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Pictory
Pictory model
Pictory model is a directory label for Pictory's video editing product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Reclaim.ai
Reclaim scheduler
Reclaim scheduler is a directory label for Reclaim.ai's calendar and task scheduling product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Replit
Replit AI
Replit AI is a directory label for Replit's interface and application generation product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Resemble
Resemble model
Resemble model is a directory label for Resemble's speech synthesis product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Runway
Runway Act-Two
Runway Act-Two is a Runway model for video generation. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Runway
Runway Aleph
Runway Aleph is a Runway model-family or version-line entry for video editing; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Runway
Runway Gen
Runway Gen is a Runway model-family or version-line entry for video generation; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Runway
Runway Gen-4
Runway Gen-4 is a Runway model for video generation. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. The official positioning emphasizes visual references for subject, object and scene consistency. This is a design goal, not a success-rate guarantee tested by this directory.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Runway
Runway Gen-4.5
Runway Gen-4.5 is a Runway model for video generation. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Rytr
Rytr AI
Rytr AI is a directory label for Rytr's writing assistance product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Scenario
Scenario custom models
Scenario custom models is a directory label for Scenario's style customization and image-asset creation product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Shengshu Technology
Vidu model
Vidu model is a Shengshu Technology model-family or version-line entry for video generation; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Sloyd
Sloyd generator
Sloyd generator is a directory label for Sloyd's 3D asset generation product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Speechify
Speechify Voice
Speechify Voice is a directory label for Speechify's speech synthesis product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Spline
Spline AI
Spline AI is a directory label for Spline's browser-based 3D design product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Stability AI
SD Turbo
SD Turbo is a Stability AI model for image generation and editing. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. The official card documents single-step text-to-image use; SD Turbo and SDXL-Turbo are separate models.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Stability AI
SDXL
SDXL is a Stability AI model for image generation and editing. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. The official card documents a base-plus-refiner pipeline; do not conflate SDXL with SD Turbo’s single-step setup.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Stability AI
Stable Audio 2.x
Stable Audio 2.x is a Stability AI model-family or version-line entry for music and sound-effect generation; exact snapshots, API availability, and licensing must be checked in the official documentation. The historical introduction documents text-to-audio and transformation of uploaded audio for music and sound effects. Current access and usage rights need separate confirmation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Stability AI
Stable Diffusion
Stable Diffusion is a Stability AI model-family or version-line entry for image generation and editing; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Stability AI
SVD（Stable Video Diffusion）
SVD（Stable Video Diffusion） is a Stability AI model for video generation. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. The cited SVD Image-to-Video model conditions video generation on a still image, rather than providing arbitrary text-to-video in the same interface.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Submagic
Submagic caption model
Submagic caption model is a directory label for Submagic's captioning and short-video editing product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Suno
Suno v5
Suno v5 is a Suno model for music and song generation. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. The official v6 FAQ says pre-v6 models are retired; retained songs do not imply that their original model can still generate.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Suno
Suno v6
Suno v6 is a Suno model for music and song generation. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Synthesia
Synthesia avatars
Synthesia avatars is a directory label for Synthesia's avatar video and voiceovers product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Tencent
Hunyuan models
Hunyuan models is a Tencent model-family or version-line entry for video generation; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Topaz Labs
Topaz Photo AI model
Topaz Photo AI model is a directory label for Topaz Labs's image upscaling and restoration product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Topaz Labs
Topaz Video model
Topaz Video model is a directory label for Topaz Labs's video enhancement product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Tripo AI / Stability AI
Tripo SR
Tripo SR is a Tripo AI / Stability AI model for single-image 3D reconstruction. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. The official repository credits Tripo AI and Stability AI jointly and specifies single-image 3D reconstruction, not arbitrary direct text-to-3D generation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
VEED
VEED AI
VEED AI is a directory label for VEED's video editing product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Vercel
v0 model
v0 model is a directory label for Vercel's interface and application generation product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Vidnoz
Vidnoz avatars
Vidnoz avatars is a directory label for Vidnoz's avatar video and voiceovers product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Voicemod
Voicemod voice model
Voicemod voice model is a directory label for Voicemod's real-time voice changing product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Wonder Dynamics
Wonder model
Wonder model is a directory label for Wonder Dynamics's markerless motion capture product capabilities, not a claim that the vendor released a standalone foundation model under this name.
Official identity and purpose checked
Availability not confirmed
Sources checked:
xAI
Grok Imagine
An image/video generation and editing family. Current docs distinguish Image 2.0, Video 1.5, and Video 1.5 Lite, including lip-synced video speech; use specific API IDs.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Zhipu AI
GLM-4 family
GLM-4 family is a Zhipu AI model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Zhipu AI
GLM-4.6
GLM-4.6 is a Zhipu AI model for reasoning, coding, and tool use. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Zhipu AI
GLM-5
GLM-5 is a Zhipu AI model for reasoning, coding, and tool use. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Sources checked:
Multiple vendors
Any LLM
A compatibility grouping for tools configurable with multiple LLM providers. Actual supported models, APIs, and permissions are tool-specific; it does not literally mean every LLM.
Directory definition
No source-check date recorded
Multiple vendors
Proprietary models
A grouping for tool-owned or undisclosed model records. It does not establish that a vendor trained a foundation model from scratch or uses no third-party components.
Directory definition
No source-check date recorded
Multiple vendors
Third-party models
A grouping by supplier relationship: tools integrating models from other providers. This is not a standalone model; exact versions and current integrations require tool-specific evidence.
Directory definition
No source-check date recorded
Open-source ecosystem
Local models
A directory grouping by deployment: inference runs on user-controlled hardware. Offline operation, telemetry, and data transfer depend on the actual software configuration.
Directory definition
No source-check date recorded
Open-source ecosystem
Open-source models
A directory grouping for open-model entries. Open weights do not automatically mean open source or unrestricted commercial use; inspect code, weights, training materials, and licenses separately.
Directory definition
No source-check date recorded
User-trained
Custom models
A grouping for user-trained, fine-tuned, or adapted models and LoRAs. Base models, training permissions, data rights, and output licenses need separate checks.
Directory definition
No source-check date recorded
Alibaba
Qwen-Max
A text model in the official legacy list, supporting tool calls and structured output; distinct from Qwen3.8-Max and vision variants.
Official identity and purpose checked
Historical source record
Sources checked:
Alibaba
Qwen-Plus
A legacy general text model still listed officially, with reasoning and tool calls; Qwen3.7-Plus is a separate model.
Official identity and purpose checked
Historical source record
Sources checked:
Anthropic
Claude Sonnet 4.5
The claude-sonnet-4-5-20250929 snapshot was deprecated on 2026-09-30 for retirement on 2026-11-30; it is not yet shut down at this review.
Official identity and purpose checked
Marked deprecated by vendor
Sources checked:
ByteDance
Seedream 4.0
The 4.0 model unifying image generation and editing with multiple references and prompt edits; the generic Seedream name does not identify this version.
Official identity and purpose checked
Historical source record
Sources checked:
DeepSeek
DeepSeek V3.2
The released V3.2 model for reasoning and agents, with tool use in thinking and non-thinking modes; distinct from V3.2-Exp and Speciale.
Official identity and purpose checked
Historical source record
Sources checked:
DeepSeek
DeepSeek V4 Flash
The original V4 Flash API model is retired; the accepted legacy ID deepseek-v4-flash now routes requests to V4.1 Flash.
Official identity and purpose checked
Historical source record
Sources checked:
Gemini 3 Pro
A historical Pro preview: gemini-3-pro-preview shut down on 2026-03-09. Google recommends Gemini 3.1 Pro Preview; the Gemini family itself remains available.
Official identity and purpose checked
Vendor service retired
Sources checked:
Google DeepMind
Veo 3
The historical Veo 3.0 line. Gemini API endpoints veo-3.0-generate-001 and veo-3.0-fast-generate-001 shut down on 2026-06-30; check other platforms separately.
Official identity and purpose checked
Vendor service retired
Sources checked:
Moonshot AI
Kimi K2
An earlier reasoning, coding and tool-use family. Hosted K2 APIs ended on 2026-05-25; the model repository is a separate weights resource.
Official identity and purpose checked
Historical source record
Sources checked:
Moonshot AI
Kimi K2.5
A visual understanding and coding model whose hosted API ended on 2026-08-31; its official repository separately documents weights and deployment.
Official identity and purpose checked
Historical source record
Sources checked:
OpenAI
DALL·E 2
A historical image generation and editing model. The dall-e-2 API shut down on 2026-05-12 and should not be recommended as callable for new projects.
Official identity and purpose checked
Vendor service retired
Sources checked:
OpenAI
DALL·E 3
A historical text-to-image model. The dall-e-3 API shut down on 2026-05-12; this record identifies the name in older projects.
Official identity and purpose checked
Vendor service retired
Sources checked:
OpenAI
GPT-5
An older text, vision, and coding model. The gpt-5-2025-08-07 snapshot is deprecated for shutdown on 2026-12-11, not the entire GPT family.
Official identity and purpose checked
Marked deprecated by vendor
Sources checked:
OpenAI
Sora 2
A historical video-with-audio model. Sora-2, sora-2-pro, and the Videos API shut down on 2026-09-24; remaining documentation does not imply availability.
Official identity and purpose checked
Vendor service retired
Sources checked:
OpenAI
Sora 2 Pro
The separate Pro video model in Sora 2. Sora-2-pro and its snapshots shut down with the Videos API on 2026-09-24; retained for identifying older configurations.
Official identity and purpose checked
Vendor service retired
Sources checked:
xAI
Grok 3
The grok-3 model retired on 2026-05-15; requests to its old ID automatically use Grok 4.3. This historical record distinguishes routing from original-model availability.
Official identity and purpose checked
Vendor service retired
Sources checked:
xAI
Grok 4
A historical Grok 4: grok-4-0709 retired on 2026-05-15. Its old ID still resolves but requests are served by Grok 4.3, not the original model.
Official identity and purpose checked
Vendor service retired
Sources checked:
xAI
Grok Imagine Image Quality
An older image model scheduled to retire on 2026-11-02, after which its ID routes to Image 2.0 at low quality. It is not yet retired on the review date.
Official identity and purpose checked
Marked deprecated by vendor
Sources checked:
Anthropic
Claude 3 Haiku
A historical Haiku: claude-3-haiku-20240307 retired on 2026-04-20, unlike still-available Haiku 4.5 and current Haiku 5.5.
Official identity and purpose checked
Vendor service retired
Directory history record
Sources checked:
Anthropic
Claude 3 Opus
A historical Opus: claude-3-opus-20240229 retired from the Claude API on 2026-01-05. Retained for old-name and migration lookup.
Official identity and purpose checked
Vendor service retired
Directory history record
Sources checked:
Anthropic
Claude 3.5 Sonnet
A historical Sonnet model: both the 20240620 and 20241022 API snapshots retired on 2025-10-28 and are no longer current endpoints.
Official identity and purpose checked
Vendor service retired
Directory history record
Sources checked:
Black Forest Labs
Flux Dev
Flux Dev is a Black Forest Labs model for image generation and editing. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. This model has published weights; read the weight license and output-use terms separately instead of calling it unrestricted open-source commercial use.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
Black Forest Labs
Flux Schnell
Flux Schnell is a Black Forest Labs model for image generation and editing. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. The official card positions it for few-step generation; schnell and dev have separate releases and license cards and are not interchangeable names.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
DeepSeek
DeepSeek V3
DeepSeek V3 is a DeepSeek model for reasoning, coding, and tool use. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
DeepSeek
DeepSeek V4 family
DeepSeek V4 family is a DeepSeek model-family or version-line entry for reasoning, coding, and tool use; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
ElevenLabs
Eleven Multilingual v2
Eleven Multilingual v2 is an ElevenLabs model for speech synthesis. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
ElevenLabs
Eleven Turbo v2
Eleven Turbo v2 is an ElevenLabs model for speech synthesis. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
Gemini Flash
Gemini Flash is a Google model-family or version-line entry for text and image understanding; exact snapshots, API availability, and licensing must be checked in the official documentation. Historical documentation positions 1.5 Flash for lightweight, high-volume work. Flash is a version-line label, not one fixed snapshot.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
Gemini Pro
Gemini Pro is a Google model-family or version-line entry for text and image understanding; exact snapshots, API availability, and licensing must be checked in the official documentation. Historical sources distinguish 1.0 Pro and 1.5 Pro. This entry groups the Pro line and does not establish a currently offered endpoint.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
Gemini Ultra
Gemini Ultra is a Google model for text and image understanding. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. The cited historical announcement identifies Gemini 1.0 Ultra for complex tasks; its historical positioning does not establish current endpoint availability.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
Imagen 4
A historical text-to-image line. Its Gemini API standard, Ultra, and Fast endpoints shut down on 2026-08-17; this does not establish availability on other cloud platforms.
Official identity and purpose checked
Vendor service retired
Directory history record
Sources checked:
Midjourney
Midjourney V5
Midjourney V5 is a Midjourney model-family or version-line entry for image generation and editing; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
Midjourney
Midjourney V6
Midjourney V6 is a Midjourney model-family or version-line entry for image generation and editing; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
OpenAI
GPT-3.5
GPT-3.5 is an OpenAI model-family or version-line entry for text understanding and generation; exact snapshots, API availability, and licensing must be checked in the official documentation.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
OpenAI
GPT-4o
An older text-and-image input, text-output model. Its 2024-08-06 and 2024-11-20 snapshots are deprecated for shutdown on 2027-01-14; audio APIs are separate.
Official identity and purpose checked
Marked deprecated by vendor
Directory history record
Sources checked:
OpenAI
Sora
Sora is an OpenAI model-family or version-line entry for video generation and editing with audio; exact snapshots, API availability, and licensing must be checked in the official documentation. This is a video-model family entry; consult the official deprecations table for Sora 2 API shutdown, rather than treating the family name as an active endpoint.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
Runway
Runway Gen-3 Alpha
Runway Gen-3 Alpha is a Runway model for video generation. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. The release describes joint image/video training and text- or image-driven video creation; it does not establish that current plans include every historical feature.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
Suno
Suno V3
Suno V3 is a Suno model for music and song generation. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. The official v6 FAQ says pre-v6 models are retired; retained songs do not imply that their original model can still generate.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
Suno
Suno V4
Suno V4 is a Suno model for music and song generation. This entry checks the documented model identity and task scope, not any third-party tool's current default configuration. The official v6 FAQ says pre-v6 models are retired; retained songs do not imply that their original model can still generate.
Official identity and purpose checked
Availability not confirmed
Directory history record
Sources checked:
Adobe
Adobe Enhance
Adobe Enhance is a directory label for Adobe's speech cleanup and enhancement product capabilities, not a claim that the vendor released a standalone foundation model under this name. Readable product evidence was unavailable in this review; the capability scope remains unverified.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Adobe
Firefly 2
Firefly 2 is retained as a Adobe directory record for image generation and editing. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Adobe
Firefly 3
Firefly 3 is retained as a Adobe directory record for image generation and editing. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Alibaba
Qwen Turbo
Qwen Turbo is retained as a Alibaba directory record for text understanding and generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Alibaba
Qwen3-Max
Qwen3-Max is retained as a Alibaba directory record for text understanding and generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Alibaba
Wan 2.x
Wan 2.x is retained as a Alibaba directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Black Forest Labs
Flux Pro
Flux Pro is retained as a Black Forest Labs directory record for image generation and editing. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
Directory history record
No source-check date recorded
Blockade Labs
Skybox model
Skybox model is a directory label for Blockade Labs's panoramic environment generation product capabilities, not a claim that the vendor released a standalone foundation model under this name. Readable product evidence was unavailable in this review; the capability scope remains unverified.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Boomy
Boomy model
Boomy model is a directory label for Boomy's music and song generation product capabilities, not a claim that the vendor released a standalone foundation model under this name. Readable product evidence was unavailable in this review; the capability scope remains unverified.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Character.AI
Character AI
Character AI is a directory label for Character.AI's conversational interaction product capabilities, not a claim that the vendor released a standalone foundation model under this name. Readable product evidence was unavailable in this review; the capability scope remains unverified.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Consensus
Consensus AI
Consensus AI is a directory label for Consensus's research discovery and question answering product capabilities, not a claim that the vendor released a standalone foundation model under this name. Readable product evidence was unavailable in this review; the capability scope remains unverified.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
CSM
CSM model
CSM model is a directory label for CSM's 3D asset generation product capabilities, not a claim that the vendor released a standalone foundation model under this name. Readable product evidence was unavailable in this review; the capability scope remains unverified.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Deemos
Rodin Gen-2
Rodin Gen-2 is retained as a Deemos directory record for 3D asset generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
ElevenLabs
Eleven Music
Eleven Music is retained as a ElevenLabs directory record for music and song generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Freepik
Mystic
Mystic is retained as a Freepik directory record for image generation and editing. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Kuaishou
Kling
Kling is retained as a Kuaishou directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Kuaishou
Kling 1.0
Kling 1.0 is retained as a Kuaishou directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
Directory history record
No source-check date recorded
Kuaishou
Kling 1.5
Kling 1.5 is retained as a Kuaishou directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
Directory history record
No source-check date recorded
Kuaishou
Kling 2.5
Kling 2.5 is retained as a Kuaishou directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Kuaishou
Kling 3.0
Kling 3.0 is retained as a Kuaishou directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
LALAL.AI
Phoenix model
Phoenix model is retained as a LALAL.AI directory record for vocal and instrument separation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Leonardo AI
Leonardo Diffusion
Leonardo Diffusion is retained as a Leonardo AI directory record for image generation and editing. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
LOVO
LOVO Genny
LOVO Genny is a directory label for LOVO's speech synthesis product capabilities, not a claim that the vendor released a standalone foundation model under this name. Readable product evidence was unavailable in this review; the capability scope remains unverified.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Luma AI
Dream Machine 1.5
Dream Machine 1.5 is retained as a Luma AI directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Luma AI
Luma Capture
Luma Capture is a directory label for Luma AI's 3D asset generation product capabilities, not a claim that the vendor released a standalone foundation model under this name. Readable product evidence was unavailable in this review; the capability scope remains unverified.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Meshy
Meshy-1
Meshy-1 is retained as a Meshy directory record for 3D asset generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Meshy
Meshy-2
Meshy-2 is retained as a Meshy directory record for 3D asset generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
MiniMax
Hailuo 2.3
Hailuo 2.3 is retained as a MiniMax directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
MiniMax
Hailuo video model
Hailuo video model is retained as a MiniMax directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
NovelAI
Kayra
Kayra is retained as a NovelAI directory record for story and fiction generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
NovelAI
NAI Diffusion
NAI Diffusion is retained as a NovelAI directory record for illustration and anime-style image generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Pika
Pika 1.0
Pika 1.0 is retained as a Pika directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Pika
Pika 2.0
Pika 2.0 is retained as a Pika directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Play.ht
PlayHT 2.0
PlayHT 2.0 is retained as a Play.ht directory record for speech synthesis. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Playground
Playground v2
Playground v2 is retained as a Playground directory record for image generation and editing. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
PrismML
Bonsai Image 4B
Bonsai Image 4B is retained as a PrismML directory record for image generation and editing. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Recraft
Recraft V3
Recraft V3 is retained as a Recraft directory record for image generation and editing. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Runway
Runway Gen-2
Runway Gen-2 is retained as a Runway directory record for video generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
Directory history record
No source-check date recorded
Stability AI
Stable Diffusion 1.5
Stable Diffusion 1.5 is retained as a Stability AI directory record for image generation and editing. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
Udio
Udio V2
Udio V2 is retained as a Udio directory record for music and song generation. The exact model label was not confirmed in this review; the official reference is not evidence of release date, active status, or a tool's configuration.
Exact identity or scope unverified
Availability not confirmed
No source-check date recorded
How to read these records
Checking identity and purpose does not establish current availability or a third-party integration. Service status follows the cited official sources; a retired service does not mean open weights disappeared. Directory history labels reflect this site's archive policy, not an official retirement decision.