Models
The table below is read live from the platform's model service and stays in sync with the console.
Copy model names directly
When calling the API, copy the value from the Model column below. Case and hyphens must match exactly — typing it by hand invites mistakes.
| Vendor | Model | Endpoints |
|---|---|---|
| Zhipu | OpenAI-compatible | |
| OpenAI-compatible | ||
| Anthropic | Anthropic, OpenAI-compatible | |
| Anthropic, OpenAI-compatible | ||
| Anthropic, OpenAI-compatible | ||
| Anthropic, OpenAI-compatible | ||
| DeepSeek | OpenAI-compatible | |
| OpenAI-compatible | ||
| Gemini, OpenAI-compatible | ||
| Gemini, OpenAI-compatible | ||
| Gemini, OpenAI-compatible | ||
| MiniMax Token Plan (minimax.io) | OpenAI-compatible | |
| OpenAI-compatible | ||
| OpenAI-compatible | ||
| Moonshot | OpenAI-compatible | |
| OpenAI-compatible | ||
| OpenAI-compatible | ||
| OpenAI-compatible | ||
| OpenAI | OpenAI-compatible | |
| OpenAI-compatible | ||
| OpenAI-compatible | ||
| OpenAI-compatible | ||
| OpenAI-compatible, Image generation | ||
| xAI | OpenAI-compatible | |
| OpenAI-compatible | ||
| OpenAI-compatible, Image generation | ||
| OpenAI-compatible, Image generation |
27 models in total.
Billing
Pricing varies by model and includes token-based, tiered and per-call billing. Input, output and other applicable rates can differ; check the corresponding model price. Charges are deducted from your paid or complimentary balance. The applicability and deduction order of balances depend on your account rules and billing records.
Per-model rates are authoritative in the console: view model pricing.
Choosing a model
| Use case | What to look for | Why |
|---|---|---|
| Everyday Q&A / general chat | Small, low-latency, low-cost models | Fast responses, suited to high-frequency conversation |
| Coding and agents | Strong reasoning and tool-calling | Handles code comprehension and multi-turn task execution |
| Long-document summarization | Long-context models | Built for large text bodies and research material |
| Structured output / JSON | Models with solid OpenAI compatibility | Good for form extraction and automated pipelines |
| Image generation / editing | Verify generation and editing separately | An image-generation listing does not establish editing support |
| Responses API | GPT-series models | Others may return not implemented |
Checklist before you commit
- Confirm the model exists: copy the name from the table above or
/v1/models— don't type it by hand. - Confirm the endpoint matches: see the
Endpointscolumn — chat uses/v1/chat/completions; Claude-style clients use/v1/messages; Gemini uses/v1beta/models/{model}:generateContent. - Confirm parameter support: add advanced parameters one at a time. Don't enable tools, JSON mode, streaming, thinking and image parameters all at once on the first try.
- Plan a fallback for production: switch to a backup model on the same protocol when the primary fails, so upstream turbulence doesn't reach your users.