Models
TokOne supports the full Claude, ChatGPT, Gemini and other model lineups (OpenAI's GPT, Codex, and Image series; Anthropic's full Claude series; Google's full Gemini series, and more). No monthly fee, no minimum spend — billed by actual token usage in real time, with prices tracking upstream official changes. See Pricing for the full rules.
Get the authoritative live list
The table below lists mainstream models only. To see which models your key can call right now, hit the endpoint:
bash
curl https://api.tokone.ai/v1/models \
-H "Authorization: Bearer $TOKONE_API_KEY"The response matches OpenAI's GET /v1/models structure.
Text / reasoning (mainstream models)
OpenAI GPT
| Model | model value | Context (in / out) |
|---|---|---|
| GPT-5.5 (default) | gpt-5.5 | 1 M / 128 K |
| GPT-5.4 | gpt-5.4 | 1 M / 128 K |
Anthropic Claude
| Model | model value | Context (in / out) |
|---|---|---|
| Claude Opus 4.7 (strongest reasoning) | claude-opus-4-7 | 1 M / 128 K |
| Claude Opus 4.6 | claude-opus-4-6 | 1 M / 128 K |
| Claude Sonnet 4.6 | claude-sonnet-4-6 | 1 M / 64 K |
Google Gemini
| Model | model value | Context (in / out) |
|---|---|---|
| Gemini Flash 3.5 | gemini-3.5-flash | 1 M / 64 K |
gpt-5.5andgpt-5.4use/v1/chat/completions, and also support/v1/responses— clients written for the Responses API (such as Codex CLI) use/v1/responsesautomatically.- Claude models use the Anthropic-compatible
/v1/messages; Gemini models use the Google-compatible/v1beta. - Models not listed above (older versions, smaller variants) are callable too; the authoritative list is
GET /v1/modelsand the console.
How to choose
| Use case | Recommended |
|---|---|
| Default / primary: long-form writing, complex reasoning, code review, agent orchestration | gpt-5.5 / claude-opus-4-7 |
| High-frequency chat, support bots, bulk translation (cost-sensitive) | gpt-4o-mini / gemini-3.5-flash |
| In-IDE autocomplete, Codex CLI, code rewriting | gpt-5.5 |
| Very long context, large-scale document processing | gemini-3.5-flash |
WARNING
Model IDs must match upstream. If a call returns 404 or "model unavailable," check GET /v1/models first to confirm the model is currently listed.