Models and rates
One price sheet for every model.
Every model below runs on the same API key and the same prepaid balance. Rates are in US dollars per one million tokens, and they are exactly what the usage log deducts.
- Models available
- 11
- Vendors
- 4
- Lowest paid input rate
- $0.24 / 1M
- API formats
- OpenAI, Anthropic
Model rate table
11 models
| Model | Input | Output | Cached | Discount | Context | Capabilities |
|---|---|---|---|---|---|---|
| Input$4.40 | Output$22.00 | Cached$0.22 | No discount | 1M | ToolsVisionReasoningStructured outputPrompt caching | |
| Input$2.20 | Output$11.00 | Cached$0.22 | No discount | 1M | ToolsVisionReasoningStructured outputPrompt caching | |
| Input$1.10 | Output$5.50 | Cached$0.11 | No discount | 200k | ToolsVisionReasoningStructured outputPrompt caching | |
| Input$2.20 | Output$11.00 | Cached$0.11 | No discount | 1.05M | ToolsVisionReasoningJSON modeStructured outputPrompt caching | |
| Input$5.50 | Output$33.00 | Cached$0.55 | No discount | 1.05M | ToolsVisionReasoningJSON modeStructured outputPrompt caching | |
| Input$2.20 | Output$13.20 | Cached$0.22 | No discount | 1.05M | ToolsVisionReasoningJSON modeStructured outputPrompt caching | |
| Input$2.20 | Output$6.60 | Cached$0.55 | No discount | 500k | ToolsVisionReasoningStructured outputPrompt caching | |
| Input$2.20 | Output$11.00 | Cached$0.22 | No discount | 1M | ToolsVisionReasoningStructured outputPrompt caching | |
| Input$2.20 | Output$6.60 | Cached$0.55 | No discount | 500k | ToolsVisionReasoningStructured outputPrompt caching | |
| Input$0.24 | Output$0.96 | Cached$0.0048 | No discount | 1M | ToolsReasoningJSON modePrompt caching | |
| Input$1.32 | Output$3.96 | Cached$0.044 | No discount | 1M | ToolsReasoningJSON modePrompt caching |
No models match these filters.
USD per 1M tokens. Struck-through rates are the regular price during a promotion. “—” in Cached means no separate cached-input rate.Stars and hidden models are saved in this browser only. The catalogue is also available as models.json.
Weighing two or three options? Compare models side by side.
Choosing a model
Pick by the job, not the name.
Start with a balanced model, move up a tier only when results fall short, and push repetitive steps down to a fast one. Switching is a one-line change to the model field.
Hardest problems Flagship
Long agentic sessions, multi-file refactors and reasoning where a wrong answer costs more than the tokens.
Everyday work Balanced
Most coding, writing and retrieval. Strong quality at a noticeably lower rate than a flagship.
Volume and sub-agents Fast
Classification, extraction, chat and helper agents, where speed and cost matter more than depth.
Two formats, one key
Use the SDK you already have.
Every model speaks the OpenAI format. Claude Opus 5.5, Claude Sonnet 5, Claude Haiku 4.5, Claude Sonnet 5.5 also accept the native Anthropic Messages format, which is what Claude Code expects.
- OpenAI formatAll 11 models
https://api.oneaiapikey.com/v1Chat Completions, Responses and Models. Header
Authorization: Bearer sk-…- Anthropic format4 Claude models
https://api.oneaiapikey.comMessages API at
/v1/messages. Headerx-api-key, orANTHROPIC_AUTH_TOKENin Claude Code.
Try any of them from a $3 top-up.
One sign-up, one key, every model in the table. Add $3 or more and pay per token; the whole deposit is yours to spend.
Model names and trademarks belong to their owners. Relaykit is an independent API gateway and is not affiliated with, endorsed by or sponsored by Anthropic, OpenAI, xAI, DeepSeek.