Models & Pricing
Supported Model Families
Overview of the AI model families available through Nyxos API and how to pick one.
Nyxos API aggregates frontier model families behind a single key. Exact model IDs rotate as providers ship new versions, so treat this page as a guide to the families — and always copy the precise model ID from the live catalog: api.nyxos.shop/pricing.
Frontier Families on the Gateway
| Family | Provider | Highlights | Recommended for |
|---|---|---|---|
| Claude | Anthropic | Strong agentic coding, long-context reasoning, prompt caching and extended thinking. | Claude Code, Cline, Roo Code, complex refactors. |
| GPT (Codex) | OpenAI | General-purpose workflows, structured output, tool calling, code generation. | Everyday coding, automation, JSON-schema pipelines. |
| Grok | xAI | Fast responses with real-time world knowledge. | Q&A, research assistance, low-latency chat. |
New models are added to the catalog regularly — the /pricing page is always the source of truth.
Choosing the Right Model
- Start from the task: agentic coding → Claude; general tooling and structured output → GPT; realtime awareness → Grok.
- Check cost: compare the Model Ratio, Completion Ratio and group multipliers on
/pricing. - Verify access: the model must list your token's group under Enabled Groups (the automatic
Auto-Agroup covers the catalog). - Test cheaply: run a short prompt with your
sk-...token and confirm output quality before wiring it into production.
Note on Model IDs
Model identifiers in API calls must match the catalog exactly (names like claude-..., gpt-..., grok-...). Provider marketing names change with each release; never hard-code an ID from memory — copy it from /pricing or GET /v1/models at runtime.