Math & Reasoning Leaderboard

Models ranked by Math Index, based on independent benchmark evaluations.

Each row shows the model's benchmark score alongside its pricing and output speed, so you can evaluate quality-to-cost tradeoffs at a glance.

Top 20 models ranked by math index

Rank	Model	Creator	Math Index	Price/1M	Speed
🥇	OpenAI: GPT-5.2 OpenAI	OpenAI	99.0	$4.81	85 tok/s
🥈	OpenAI: GPT-5 Codex OpenAI	OpenAI	98.7	$3.44	178 tok/s
🥉	Gemini 3 Flash Preview (Reasoning) Google	Google	97.0	$1.13	215 tok/s
4	GPT-5.2 (medium) OpenAI	OpenAI	96.7	$4.81	—
5	DeepSeek V3.2 Speciale DeepSeek	DeepSeek	96.7	$0.32	—
6	MiMo-V2-Flash (Reasoning) Xiaomi	Xiaomi	96.3	$0.15	91 tok/s
7	Gemini 3 Pro Preview (high) Google	Google	95.7	$4.50	161 tok/s
8	OpenAI: GPT-5.1-Codex OpenAI	OpenAI	95.7	$3.44	206 tok/s
9	Z.ai: GLM 4.7 Z AI	Z AI	95.0	$0.74	119 tok/s
10	KAT-Coder-Pro V1 KwaiKAT	KwaiKAT	94.7	$0.53	109 tok/s
11	MoonshotAI: Kimi K2 Thinking Kimi	Kimi	94.7	$1.07	132 tok/s
12	OpenAI: GPT-5 OpenAI	OpenAI	94.3	$3.44	99 tok/s
13	Nova 2.0 Lite (high) Amazon	Amazon	94.3	$0.85	188 tok/s
14	OpenAI: GPT-5.1 OpenAI	OpenAI	94.0	$3.44	95 tok/s
15	OpenAI: gpt-oss-120b OpenAI	OpenAI	93.4	$0.06	307 tok/s
16	Grok 4 xAI	xAI	92.7	$11.00	47 tok/s
17	DeepSeek V3.2 (Reasoning) DeepSeek	DeepSeek	92.0	$0.34	79 tok/s
18	GPT-5 (medium) OpenAI	OpenAI	91.7	$3.44	97 tok/s
19	OpenAI: GPT-5.1-Codex-Mini OpenAI	OpenAI	91.7	$0.69	216 tok/s
20	Claude Opus 4.5 (Reasoning) Anthropic	Anthropic	91.3	$10.00	57 tok/s