POST /v1/chat/completions
OpenAI Chat Completions
client.chat.completions.create( model="claude-fable-5", messages=[...], )
OpenAI + Anthropic compatible · 43 models · 10 model makers
One key. 43 models. Any agent.
Point Claude Code, OpenCode, Codex, or any SDK at one endpoint. The router picks a healthy provider at the best cost and keeps your session on the same route so the prompt cache keeps hitting.
# Install AivoCode (aivo) v1.0.1 & authenticate
npm install -g --force https://aivokey.dev/cli/download/aivocode-1.0.1.tgz
aivo login sk-aivo_…
aivo --model claude-fable-5/models
The catalog, priced per 1M tokens
The same data GET /v1/models returns: context window, capabilities and cache pricing.
cheapest input first
- MiMo V2.5−65%
mimo-v2.5
- input
- $0.049
- output
- $0.098
- context
- 1M
- GLM 5.3 Flash−65%
glm-5.3-flash
- input
- $0.0525
- output
- $0.175
- context
- 1M
- Qwen3.8 Omni Flash−65%
qwen3.8-omni-flash
- input
- $0.0525
- output
- $0.1645
- context
- 1M
- GPT 5.6 Luna−65%
gpt-5.6-luna
- input
- $0.07
- output
- $0.42
- context
- 1.05M
- GPT 6 Luna−65%
gpt-6-luna
- input
- $0.07
- output
- $0.2625
- context
- 1.05M
- Qwen3.8 27B−65%
qwen3.8-27b
- input
- $0.07
- output
- $0.28
- context
- 1M
| Model | Maker | Context | Input ↑ | Output | Capabilities |
|---|---|---|---|---|---|
| MiMo V2.5 mimo-v2.5 | Xiaomi | 1M | $0.049 −65% | $0.098 | vision · tools · reasoning · cache · audio |
| GLM 5.3 Flash glm-5.3-flash | Z.ai | 1M | $0.0525 −65% | $0.175 | vision · tools · reasoning · cache · file |
| Qwen3.8 Omni Flash qwen3.8-omni-flash | Alibaba Qwen | 1M | $0.0525 −65% | $0.1645 | vision · tools · reasoning · cache |
| GPT 5.6 Luna gpt-5.6-luna | OpenAI | 1.05M | $0.07 −65% | $0.42 | vision · tools · reasoning · cache |
| GPT 6 Luna gpt-6-luna | OpenAI | 1.05M | $0.07 −65% | $0.2625 | vision · tools · reasoning · cache |
| Qwen3.8 27B qwen3.8-27b | Alibaba Qwen | 1M | $0.07 −65% | $0.28 | vision · cache |
/v1
Three API formats, one key
Any model through any format. Send the key as Authorization: Bearer or x-api-key.
POST /v1/responses
OpenAI Responses
client.responses.create( model="claude-fable-5", input="…", )
POST /v1/messages
Anthropic Messages
client.messages.create( model="claude-fable-5", max_tokens=1024, messages=[...], )
routing
What happens to one request
01
Your agent sends it
Claude Code, OpenCode, Codex, or an SDK, with the model's real name.
02
Key and budget are checked
The cost is reserved first, then settled from actual usage.
03
A route is chosen
Provider cost, health and capabilities. Your session stays on the same route.
04
Served, or rerouted
A failing route is cooled down and the request moves to another healthy provider. Provider names are never exposed.
budget
A fixed order of spend
Each request is paid from the first source that still has funds. When an allowance runs out you get a machine-readable error and the reset time.
01bonus tokensfrom your plan · expire at period end
02free tokensclaimed per model on the faucet
03plan allowanceper model or pooled · 5h / 7d / monthly
04balanceper token · fallback is per key
HTTP 402
"error_code": "model_allowance_exhausted",
"window": "5h",
"reset_at": "…"❯ Plans from $0.56/month