LLM Leaderboard

Ranked by LMArena Elo · source LMArena

Human preference Elo from chatbot arena head-to-head votes.

IntelligenceAA LMArenaLMArena CodingAA MathAA MMLU-ProAA GPQAAA LiveCodeAA MATH-500AA AIMEAA HLEAA SciCodeAA

#1 by LMArena Elo

GPT-5.5 (high)

1469

OpenAI · in $5.00 / out $30.0 per 1M · 79 tok/s

#	Model	Creator	LMArena	In $/1M	Out $/1M	tok/s	LMArena
1	GPT-5.5 (high)	OpenAI	1469	$5.00	$30.0	79	1469
2	GPT-5.5 (xhigh)	OpenAI	1464	$5.00	$30.0	86	1464
3	GPT-5.5 (medium)	OpenAI	1464	$5.00	$30.0	77	1464
4	GPT-5.5 (low)	OpenAI	1464	$5.00	$30.0	79	1464
5	GPT-5.5 (Non-reasoning)	OpenAI	1464	$5.00	$30.0	74	1464
6	GPT-5.5 Instant (May 2026)	OpenAI	1464	$5.00	$30.0	0	1464
7	GPT-5.5 Instant (June 2026)	OpenAI	1464	$5.00	$30.0	0	1464
8	GPT-5.5 Pro (xhigh)	OpenAI	1464	$0.00	$0.00	0	1464
9	Gemini 2.5 Pro	Google	1462	$1.25	$10.0	151	1462
10	Gemini 2.5 Pro Preview (Mar' 25)	Google	1462	$0.00	$0.00	0	1462
11	Gemini 2.5 Pro Preview (May' 25)	Google	1462	$1.25	$10.0	0	1462
12	GPT-5.4 (xhigh)	OpenAI	1462	$2.50	$15.0	163	1462
13	GPT-5.4 mini (xhigh)	OpenAI	1462	$0.75	$4.50	180	1462
14	GPT-5.4 (low)	OpenAI	1462	$2.50	$15.0	92	1462
15	GPT-5.4 nano (xhigh)	OpenAI	1462	$0.20	$1.25	167	1462
16	GPT-5.4 nano (medium)	OpenAI	1462	$0.20	$1.25	174	1462
17	GPT-5.4 mini (medium)	OpenAI	1462	$0.75	$4.50	203	1462
18	GPT-5.4 (Non-reasoning)	OpenAI	1462	$2.50	$15.0	103	1462
19	GPT-5.4 nano (Non-Reasoning)	OpenAI	1462	$0.20	$1.25	168	1462
20	GPT-5.4 mini (Non-Reasoning)	OpenAI	1462	$0.75	$4.50	168	1462
21	GPT-5.4 Pro (xhigh)	OpenAI	1462	$30.0	$180.0	0	1462
22	Grok 4.5 (high)	SpaceXAI	1460	$2.00	$6.00	75	1460
23	Grok 4	SpaceXAI	1460	$5.50	$27.5	0	1460
24	Gemini 3 Pro Preview (high)	Google	1457	$2.00	$12.0	0	1457
25	Gemini 3 Pro Preview (low)	Google	1457	$2.00	$12.0	0	1457
26	GPT-5.6 Sol (xhigh)	OpenAI	1457	$5.00	$30.0	65	1457
27	Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)	Anthropic	1457	$10.0	$50.0	70	1457
28	Inkling (xhigh)	Thinking Machines	1454	$1.87	$4.68	0	1454
29	Claude Opus 4.8 (Adaptive Reasoning, Max Effort)	Anthropic	1453	$5.00	$25.0	62	1453
30	Claude Opus 4.6 (Adaptive Reasoning, Max Effort)	Anthropic	1449	$5.00	$25.0	64	1449
31	Claude Opus 4.6 (Non-reasoning, High Effort)	Anthropic	1449	$5.00	$25.0	51	1449
32	GLM-4.6 (Reasoning)	Z AI	1448	$0.55	$2.20	46	1448
33	GLM-4.6 (Non-reasoning)	Z AI	1448	$0.57	$2.20	68	1448
34	GLM-4.6V (Reasoning)	Z AI	1448	$0.30	$0.90	65	1448
35	GLM-4.6V (Non-reasoning)	Z AI	1448	$0.30	$0.90	56	1448
36	Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)	Anthropic	1447	$3.00	$15.0	70	1447
37	Claude Sonnet 4.6 (Non-reasoning, High Effort)	Anthropic	1447	$3.00	$15.0	58	1447
38	Claude Sonnet 4.6 (Non-reasoning, Low Effort)	Anthropic	1447	$3.00	$15.0	59	1447
39	Gemini 3 Flash Preview (Reasoning)	Google	1446	$0.50	$3.00	237	1446
40	Gemini 3 Flash Preview (Non-reasoning)	Google	1446	$0.50	$3.00	221	1446
41	Grok 4.1 Fast (Reasoning)	SpaceXAI	1446	$0.00	$0.00	0	1446
42	Grok 4.1 Fast (Non-reasoning)	SpaceXAI	1446	$0.00	$0.00	0	1446
43	MiMo-V2.5-Pro	Xiaomi	1445	$0.43	$0.87	59	1445
44	MiMo-V2.5-Pro (Non-reasoning)	Xiaomi	1445	$0.66	$1.72	63	1445
45	Gemini 3.1 Pro Preview	Google	1445	$2.00	$12.0	137	1445
46	Muse Spark	Meta	1445	$0.00	$0.00	0	1445
47	GPT-5.2 (xhigh)	OpenAI	1443	$1.75	$14.0	85	1443
48	GPT-5.2 Codex (xhigh)	OpenAI	1443	$1.75	$14.0	163	1443
49	GPT-5.2 (medium)	OpenAI	1443	$1.75	$14.0	0	1443
50	GPT-5.2 (Non-reasoning)	OpenAI	1443	$1.75	$14.0	72	1443

Quality, pricing, and speed benchmarks via Artificial Analysis. Human preference Elo via LMArena. Refreshed daily at 06:00 UTC · last sync Tue, 21 Jul 2026 06:00:01 GMT.

About the metrics

Intelligence Index

Artificial Analysis composite of reasoning, knowledge, coding, math.

LMArena Elo

LMArena

Human preference Elo from chatbot arena head-to-head votes.

Coding Index

AA composite — LiveCodeBench, SciCode, etc.

Math Index

AA composite — MATH-500, AIME, etc.

MMLU-Pro

Multi-domain knowledge & reasoning. Harder than MMLU.

GPQA Diamond

Graduate-level science Q&A. Strong reasoning signal.

LiveCodeBench

Competitive programming, contamination-resistant.

MATH-500

Subset of the MATH benchmark.

AIME

American Invitational Mathematics Examination — olympiad math.

Humanity's Last Exam

Frontier reasoning benchmark; most models still score low.

SciCode

Scientific coding tasks.