Model catalog
One account, one API, every top model — Claude, Grok and GPT side by side. Switch per message in the app, name a model in an API call, or let the automatic router choose. Billed per token; no subscription.
Automatic router
Send "model": "auto" (or pick Auto in the chat) and vexer routes each request to a model that fits it — deeper reasoning for hard or long or code-heavy prompts, the fast default for everything else. Every model also has a fallback: if one is unavailable, the request reroutes before a token is streamed, so a provider blip doesn't become your error.
| Model | Family | Context | Best at | Vision | Price / 1M tokens |
|---|---|---|---|---|---|
Claude Opus 4.8claude-opus-4-8 | Anthropic | 1M | Hardest reasoning, long agentic work, coding | Yes | $15 in · $75 out |
Claude Sonnet 5claude-sonnet-5 | Anthropic | 1M | Fast, balanced default for chat and code | Yes | $3 in · $15 out |
Claude Fable 5claude-fable-5 | Anthropic | 1M | Creative writing, tone, long-form prose | Yes | $3 in · $15 out* |
Grok 4.5grok-4.5 | xAI | 256K | Current-events knowledge, casual reasoning | — | $3 in · $15 out* |
GPT-5.6 Solgpt-5.6-sol | OpenAI | 400K | Broad general-purpose, tools, structured output | Yes | $1.25 in · $10 out* |
Prices are what vexer charges, in US dollars per million tokens, billed per token against your balance. * provisional until the provider publishes an official rate. See the API docs for endpoints, or read what LLM routing is and when it helps.
Compare: Claude vs GPT vs Grok · One chat for every AI model