MindsHub Inference

Every major model, through the three APIs you already use, on one key and one bill.

Call Claude, GPT, Gemini, Kimi, DeepSeek, Qwen, Grok and more from a single endpoint, using whichever request format your code already speaks:

API	Endpoint	Works with
OpenAI Chat Completions	`POST /v1/chat/completions`	OpenAI SDKs, most of the ecosystem
OpenAI Responses	`POST /v1/responses`	OpenAI SDKs, Codex (mid-upgrade; status)
Anthropic Messages	`POST /v1/messages`	Anthropic SDKs, Claude Code

Same key, same models, same features through all three. Any model works behind any API: drive Claude Fable from the OpenAI SDK, or GPT 5.6 Sol from the Anthropic SDK, and neither client knows the difference.

Start here

Getting started

Key to working app in fifteen minutes: three APIs, streaming, tools, images, web search.

Choosing an API

Completions, Responses, or Messages, and what each supports.

Coding agents

Run Claude Code on any catalog model. Codex support is coming.

Models

The catalog, aliases, and reasoning effort levels.

The one-minute version

Base URL: https://api.mindshub.ai/v1

curl https://api.mindshub.ai/v1/chat/completions \
  -H "Authorization: Bearer $MINDSHUB_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mindshub_air",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Get a key at console.mindshub.ai. New organizations get a monthly allowance of included tokens on mindshub_air, so your first calls cost nothing.

Why teams use it

Keep your SDK. No MindsHub SDK to learn, no wrapper to install. Point the official openai or @anthropic-ai/sdk client at our base URL and your existing code runs.

Switch models by editing one string. "model": "sonnet" becomes "model": "kimi". No new client, no new credential, no re-plumbing. Benchmark Claude Opus 5 against DeepSeek V4 Pro on your real prompts in an afternoon.

One bill for everything. Every model draws the same prepaid balance and lands in one usage summary, coding agents included. No separate vendor accounts to fund, and no month-end reconciliation.

Requests aren't rejected over parameters. Models disagree about what they accept. Rather than 400-ing because one model dislikes top_k, MindsHub drops or clamps what the target model can't take and tells you in a response header. How that works.

Your coding agents come along. Claude Code points at MindsHub with two environment variables and runs on any catalog model, Kimi K3 included, billed alongside everything else. Codex support is coming.