Models Catalog
Browse every model, with pricing, context limits, capabilities and how to use each one (your own key, credits, free tier or the shared pool).
The Models page lists every model AnyRouter offers, with pricing, limits and capabilities. Open it at /models or /dashboard/models.
Everyone sees every model, signed in or not. The Access column says how to use each one: BYOK (works with your own provider key), Pool (the shared donated-key pool can serve it), or a price when credits pay for it. A model you cannot call yet is still listed; the label tells you what it needs, such as your own key. GET /api/v1/models follows the same rule and adds can_route, which says whether you can call the model right now.
Fields
| Field | Meaning |
|---|---|
| Model id | The string for the model field, for example openai/gpt-5.4-mini or z-ai/glm-4.7-flash |
| Owner | The company that publishes the model |
| Context window | Maximum prompt plus response tokens per request |
| Max output | Maximum tokens per response; a higher max_tokens is clamped |
| Pricing | US dollars per 1M input and output tokens; free-tier models show Free |
Prices are all-in, with no per-request fees. For example, 12,000 prompt and 800 output tokens at 1.00 / 4.00 cost about $0.0152. Cache-hit and cache-write rates apply where supported (see Prompt Caching).
Capability chips show what a model supports: tools, vision, structured output (response_format json_schema), streaming, prompt caching and multilingual. Missing ones are grayed out.
Privacy badges show the serving provider's tier: Zero-retention, Short-retention (no training on your data) or Standard. They reflect the upstream's public commitments and are advisory; use Privacy Controls to block providers.
Active in your workspace narrows the list to models used in the last 30 days: routed from one of your keys, referenced by a preset, or set as an API key default. It updates within minutes of a request.
Ids you can send
- Canonical id from the catalog.
- Aliases: shorter names that resolve to a canonical id.
- Virtual aliases:
anyrouter/auto,anyrouter/latest,anyrouter/freeresolve at request time to a chain of models (see Smart Routing).GET /api/v1/modelsandGET /api/v1/models/{id}return amembersarray for these:{ id, name, owner, release_date, position }, in the order a request tries them. - Presets:
@preset/<name>, optionally with a model override:openai/gpt-5.4-mini@preset/code-reviewer. Presets are listed on the Presets page, not in the catalog. - Provider preferences: a
providerobject in the request body pins an upstream, sorts by price or latency, or sets a max price.
Try a model
Use the Copy action on a row, then:
curl https://anyrouter.dev/api/v1/chat/completions \
-H "Authorization: Bearer sk-ar-your-key" \
-H "Content-Type: application/json" \
-d '{"model": "openai/gpt-5.4-mini", "messages": [{"role": "user", "content": "Reply with the single word: ready."}]}'To read the catalog in code, call GET https://anyrouter.dev/api/v1/models with your key. It is OpenAI-compatible and returns id, owner, context window, pricing and capability flags for every model, plus access and can_route for the key you send. The catalog changes rarely, so caching it for a few minutes is fine. See the Models API reference.
Troubleshooting
| Symptom | Fix |
|---|---|
| Expected model missing | Every listed model shows. Only your privacy settings and your key's own allow or block lists hide a model. Check those, then Logs if it is still missing. |
| Model misbehaves | Open Logs for the request id to see the upstream and error code. |
| Price changes | Announced in advance and recorded in Credits history; the catalog shows the current rate. |
AnyRouter routes requests to third-party models. Models and outputs are provided as is, with no warranty; you use them at your own risk. Disclaimer