Models
Model details: API IDs, context windows and availability
The technical facts about the 21 models in Milly Lab: the ID each one has in the API at https://api.m-lab.app/v1 and in error messages, its context window, where it runs — the app, the API or both — and what answers a retired ID.
Chat models: IDs and context windows
All 13 chat models with the ID the API and error messages use, the context window the API publishes and where the app offers each one.
| Model | Developer | API ID | Notes |
|---|---|---|---|
| Luna · GPT-6 | OpenAI | gpt-5.4-nano | 1.05M-token context |
| Terra · GPT-6.1 | OpenAI | gpt-5.4 | 1.05M-token context · Reads images |
| Sol · GPT-6.1 | OpenAI | gpt-5.5 | 1.05M-token context · Reads images · In the app: Swarm Mind only |
| Sol · GPT-6 | OpenAI | gpt-5.3-codex | 1.05M-token context · Reads images · In the app: Swarm Mind only |
| Claude Haiku 5.5 | Anthropic | claude-haiku-5-5 | 1M-token context · Reads images |
| Claude Sonnet 5.5 | Anthropic | claude-sonnet-5 | 1M-token context · Reads images |
| Claude Opus 5.5 | Anthropic | claude-opus-4-8 | 1M-token context · Reads images · In the app: Swarm Mind only |
| DeepSeek Flash | DeepSeek | deepseek-v4-flash | 1M-token context |
| DeepSeek Flash · Pro | DeepSeek | deepseek-v4 | 1M-token context |
| DeepSeek V4 Pro | DeepSeek | deepseek-v4-pro | 1M-token context |
| Gemini 3.8 Flash 200K | gemini-3.6-flash-200k | 200K-token context · Reads images | |
| Gemini 3.8 Flash | gemini-3.6-flash | 1.05M-token context · Reads images | |
| Grok 4.7 | xAI | grok-4.3 | 500K-token context · Not in any plan |
Image, video and 3D models
Image models run in Light Studio and through the API; video and 3D models run only in the app’s studios.
| Model | Type | Developer | API ID | Notes |
|---|---|---|---|---|
| Gemini Nano Banana 2.1 | Image | gemini-nano-banana | Light Studio and API | |
| GPT Image 2.5 Flare | Image | OpenAI | gpt-image-2 | Light Studio and API |
| Flux 3 Image | Image | Black Forest Labs via fal.ai | fal-ai/flux-pro | Light Studio and API |
| Stable Diffusion 3.5 Large | Image | Stability AI via fal.ai | fal-ai/stable-diffusion-xl | Light Studio and API |
| Gemini Omni 1.2 Flash | Video | — | Video Studio only, not in the API | |
| Kling 3.0 | Video | Kuaishou via fal.ai | — | Video Studio only, not in the API |
| Runway Gen-4 Turbo | Video | Runway | — | Video Studio only, not in the API |
| Meshy | 3D | Meshy | — | 3D Studio only, not in the API |
Which models can I use through the API?
The OpenAI-compatible Milly Lab API at https://api.m-lab.app/v1 serves the chat models (13) and image models (4) by the API IDs in the tables above, plus the text-embedding-3-small embeddings model. Video and 3D models are available only in the app.
API access is included in the Max, Advanced and Enterprise plans. GET https://api.m-lab.app/v1/models returns the live model list with prices and needs no API key.
Why do API IDs differ from model names?
API IDs are permanent: when a model is replaced by a newer version, its ID stays the same so existing code keeps working, and only the name shown in the app changes. That is why Luna · GPT-6 is called with the ID gpt-5.4-nano and Claude Opus 5.5 with claude-opus-4-8. The tables show both exactly as the platform uses them.
Which old model IDs still work in the API?
A retired API ID that has a successor still works: the request is answered and billed as the current model. An ID without a successor returns a 404 model_not_found error.
| Old API ID | Answered by |
|---|---|
claude-haiku-4-5 | claude-haiku-5-5 · Claude Haiku 5.5 |
claude-sonnet | claude-sonnet-5 · Claude Sonnet 5.5 |
claude-sonnet-4-6 | claude-sonnet-5 · Claude Sonnet 5.5 |
claude-opus | claude-opus-4-8 · Claude Opus 5.5 |
gemini-2.5-flash | gemini-3.6-flash · Gemini 3.8 Flash |
gemini-2.5-pro | gemini-3.6-flash · Gemini 3.8 Flash |
gemini-3.5-flash | gemini-3.6-flash · Gemini 3.8 Flash |
gemini-2.5-pro-200k | gemini-3.6-flash-200k · Gemini 3.8 Flash 200K |
gemini-3.5-flash-200k | gemini-3.6-flash-200k · Gemini 3.8 Flash 200K |
gpt-5.4-mini | Not served (404) |
gpt-4o | gpt-5.4-nano · Luna · GPT-6 |
gpt-4o-mini | gpt-5.4-nano · Luna · GPT-6 |
dall-e-3 | Not served (404) |
Is voice transcription included?
Voice transcription is included in Starter, Pro, Max and Advanced, with its own monthly allowance of voice notes: Starter 133, Pro 222, Max 333 and Advanced 666.
How much of each model does a plan include?
Every plan’s monthly allowance for each model — in messages, images, video clips or 3D models — is on the Models page and, plan by plan, in the Plans, billing and usage guide.