Inference Space

Performance

Latency, token rates, recent traffic, and availability from real calls — not eval scores.

ModelLatency (p50)Input token rateOutput token rate7d volume7d uptimeDetail
Grok 4.5Grok8,007ms3,778token/s23token/s20,274100.00%
GPT-5.6 LunaGPT10,012ms5,266token/s28token/s43,63398.01%
GPT-5.6 SolGPT11,693ms8,406token/s29token/s44,70371.79%
Nano Banana 2Gemini Image15,289ms29token/s27token/s20,46188.24%
GPT-5.6 TerraGPT22,538ms5,881token/s21token/s23,54597.52%
GPT Image 2GPT Image 230,206ms24token/s29token/s21,22085.11%
Nano Banana ProGemini Image53,810ms24token/s28token/s20,69279.31%
Claude Fable 5Claudeneeds metrics100.00%
Claude Haiku 4.5Claudeneeds metrics100.00%
Claude Haiku 4.5Claudeneeds metrics100.00%
Claude Opus 5Claudeneeds metrics100.00%
Claude Sonnet 5Claudeneeds metrics100.00%
GPT-5.3 CodexGPTneeds metrics100.00%
GPT-5.4GPTneeds metrics100.00%
GPT-5.4 MiniGPTneeds metrics100.00%
GPT-5.5GPTneeds metrics100.00%
grok-4.6Grokneeds metrics100.00%