Provides OpenAI-compatible API access to open-source AI models like DeepSeek and QwQ, powered by shared idle computing resources.

Suanli API offers an OpenAI-compatible interface for integrating AI models into projects. It supports models such as DeepSeek-R1 and QwQ, with endpoints like /v1/chat/completions. The service is built on shared idle computing resources from personal devices (e.g., 4090 GPUs) and institutions, aiming to provide accessible AI capabilities. Typical use cases include AI-powered chat applications and leveraging open-source models for development. The API uses standard authentication with API keys and maintains compatibility with OpenAI's parameters and response formats.

Time	Model	Speed	Latency
Dec 7, 10:31 AM	free:Qwen3-30B-A3B	30.96 t/s	10.18s
Dec 7, 10:28 AM	free:Qwen3-30B-A3B	26.84 t/s	36.04s
Dec 7, 06:12 AM	free:Qwen3-30B-A3B	16.91 t/s	0.69s
Dec 7, 03:10 AM	free:Qwen3-30B-A3B	23.40 t/s	0.98s
Dec 7, 02:35 AM	free:Qwen3-30B-A3B	28.30 t/s	10.90s
Dec 5, 09:39 AM	free:Qwen3-30B-A3B	14.89 t/s	24.83s
Nov 28, 08:10 PM	free:Qwen3-30B-A3B	19.15 t/s	17.83s
Nov 28, 12:54 PM	free:Qwen3-30B-A3B	21.40 t/s	10.58s
Nov 28, 12:25 AM	free:Qwen3-30B-A3B	12.32 t/s	8.02s
Nov 27, 04:13 PM	Qwen/Qwen3-VL-32B-Thinking	32.58 t/s	20.47s

算了么 API

Supported Models

Recent Test Records

算了么 API

About 算了么 API

Supported Models

Recent Test Records