Model Benchmarks

Full leaderboard across 11 models. Click any column header to sort.

Rank	Model	Provider	Quality ↓	Speed	Cost	+Refrase
1	Claude Sonnet 4.6	Anthropic	94	1200ms	$$$	+12%
2	Gemini Pro	Google	93	900ms	$$	—
3	GPT-4o	OpenAI	91	800ms	$$$	+8%
4	Qwen3 235B	Alibaba	88	1500ms	$$	+15%
5	DeepSeek V3	DeepSeek	87	1100ms	$	+10%
6	Mistral Large 3	Mistral	86	1000ms	$$	+7%
7	Llama 3.1 405B	Meta	85	1300ms	$$	+9%
8	Kimi K2	Moonshot	84	1400ms	$	+11%
9	GLM 4.7	Z.AI	83	900ms	$	+13%
10	Nemotron 30B	NVIDIA	82	700ms	$	+6%
11	MiniMax M2	MiniMax	81	1000ms	$	+8%