小水管

A unified gateway providing access to multiple AI models through a single API endpoint for personal and research use.

Qwen3 Max GPT-5.2 GPT-5.3 Codex GPT-5.3 Grok 4 20 Qwen3.5 Plus Kimi K2.5

小水管 offers 207 LLM API models.

Speed benchmark average: 69 tok/s.

Avg Speed69.26 tok/s

Latency6.78 s

Total Tests55

Models9

Updated4/17/2026

Created At3/4/2026

Website

API Endpoints

edge.pieixan.icu
proxy.pieixan.icu

Claim this provider

氕Verified by 氕氙

Supported Models

Model	Speed	Latency	Tests
qwen3-coder-flash	110.06 tok/s	1.19s	10
doubao-seed-2.0-pro	81.01 tok/s	22.47s	5
qwen3-max-preview	74.20 tok/s	0.60s	5
gpt-5.2-chat	71.48 tok/s	2.89s	5
gpt-5.3-codex	64.15 tok/s	2.75s	5
gpt-5.3-chat	63.77 tok/s	2.64s	10
grok-4.20-beta	54.81 tok/s	1.17s	5
qwen3.5-plus	54.49 tok/s	4.37s	5
kimi-k2.5	14.05 tok/s	32.71s	5

Available Models

198 models available (last updated: 4/18/2026)

all-mpnet-base-v2customarctic-embed-lcustombaai/bge-m3custombaai/bge-reranker-v2-m3custombig-picklecustomcb/llama3.1-8bcustomclaude-haiku-4-5-20251001vertex-aiclaude-sonnet-4-6vertex-aideepseek-chatdeepseekdeepseek-reasonerdeepseekdeepseek/deepseek-r1-0528-qwen3-8bcustomdeepseek/deepseek-v3.1-terminuscustomdeepseek/deepseek-v3.2customdevstral-2-123b-instruct-2512customdoubao-1.5-procustomdoubao-seed-1.6customdoubao-seed-1.6-flashcustomdoubao-seed-2.0-codecustomdoubao-seed-2.0-litecustomdoubao-seed-2.0-procustom

Showing 20 of 198 models.