Text & Chat Models
Choose a text model and call it through an OpenAI-, Claude-, or Gemini-compatible endpoint. The IDs below are examples; call `/v1/models` for the models available to your key.
How to choose
Compare models by the capabilities your workload needs, such as reasoning, tool use, multimodal input, context window, throughput, and structured output. Confirm availability and pricing on the Models page, then test the selected model with representative inputs before production use.
Representative model IDs
These examples show the provider/model ID format and may change as the catalog evolves. Use GET /v1/models before saving a model in application configuration.
| Parameter | Type | Required | Description |
|---|---|---|---|
| anthropic/claude-opus-4.8 | Anthropic | optional | Advanced reasoning, agentic coding, and long-context analysis. |
| openai/gpt-5.5 | OpenAI | optional | General-purpose flagship with strong tool use and structured output. |
| google/gemini-3.5-flash | optional | Fast multimodal model with a very large context window. | |
| deepseek/deepseek-v4-pro | DeepSeek | optional | Reasoning, coding, and long-form output. |
| qwen/qwen3.7-max | Qwen | optional | Large multilingual model with broad capability coverage. |
| x-ai/grok-4.3 | xAI | optional | Conversational flagship with real-time leanings. |
| moonshotai/kimi-k2.6 | MoonshotAI | optional | Long-context model strong at agentic workflows. |
| z-ai/glm-5 | Z.ai | optional | Capable bilingual (EN/ZH) general model. |
High-throughput options
| Parameter | Type | Required | Description |
|---|---|---|---|
| google/gemini-3.1-flash-lite | optional | High-throughput multimodal processing for simpler tasks. | |
| openai/gpt-5.4-nano | OpenAI | optional | Compact GPT tier for responsive workloads. |
| deepseek/deepseek-v4-flash | DeepSeek | optional | Responsive reasoning-oriented tier. |
| qwen/qwen3.6-flash | Qwen | optional | Lightweight multilingual tier. |
Specialized: coding & reasoning
Coding-tuned and reasoning models (e.g. openai/gpt-5.3-codex, qwen/qwen3-coder, moonshotai/kimi-k2-thinking) expose extra controls like reasoning_effort. See the Reasoning and Tool Calling guides.
Model-dependent features
OneRouter keeps protocol compatibility and forwards model-specific parameters where supported. Features such as tool calling, structured output, reasoning, prompt caching, web search, image input, audio input, video generation, and provider-specific options depend on the selected model and upstream provider. Check the model detail page before relying on a feature in production.