1,000 output tokens

One trip to the other side = 1,000 output tokens · A round trip = 2,000

GPT-5.6 SolCerebras CS-4 · 1,350 t/s

Undisclosedparams ↗

gpt-oss-20B (high)Groq · 949.5 t/s

20Bparams ↗

Nemotron 3.5 LightningDeepInfra (NVFP4) · 516.2 t/s

31.6B / 3.6Bparams ↗

Gemini 3.7 Flash (high)Google AI Studio · 340.1 t/s

Undisclosedparams ↗

DeepSeek V4 Flash (Non-reasoning)Makora · 274.1 t/s

284B / 13Bparams ↗

MiniMax-M3Modular · 231.5 t/s

428B / 23Bparams ↗

Muse Spark 1.2 (xhigh)Meta · ~217.2 t/s

Undisclosedparams ↗

GLM-5.2 (max)Baseten (FAST) · 195.5 t/s

744Bparams ↗

Gemini 3.5 Flash (high)Google AI Studio · 172.9 t/s

Undisclosedparams ↗

GPT-5.6 Luna (max)OpenAI · 150.2 t/s

Undisclosedparams ↗

Claude Sonnet 5 (Adaptive Reasoning, Max Effort)Google · 81.8 t/s

Undisclosedparams ↗

Grok 4.6 (high)SpaceXAI · 65.5 t/s

Undisclosedparams ↗

Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)Anthropic · 63.2 t/s

Undisclosedparams ↗

GPT-5.6 Sol (max)OpenAI · 61.7 t/s

Undisclosedparams ↗

Claude Opus 5 (Adaptive Reasoning, Max Effort)Anthropic · 54.1 t/s

Undisclosedparams ↗

Kimi K3 (max)Fireworks · 48.9 t/s

2.8Tparams ↗

Qwen3.8 MaxAlibaba Cloud · 47.0 t/s

Undisclosedparams ↗