BenchLeader
Menu
LeaderboardsModelsProvidersBenchmarksCompareReleases1BlogMethodology

LMArena Maths

Text arena rating on maths prompts.

As of 19 Sept 2026, Claude Fable 5 leads LMArena Maths on BenchLeader with 1526, ahead of Claude Opus 5 at 1525, across 351 model configurations with a published result.

Published by
LMArena
Category
Maths
Index weight
1.0
Models
351
Data as of
19 Sept 2026

CC BY 4.0 — lmarena-ai/leaderboard-dataset on Hugging Face.

What the test looks like

Pairwise votes on prompts classified as mathematics.

How it is scored

Bradley-Terry rating from pairwise votes, in Elo-like units, with style control.

What to keep in mind

Voters judge the answer as written; correctness is not verified. Counts toward the maths category since index v1.2.

Top 15
  1. 1Claude Fable 51526
  2. 2Claude Opus 5 (high)1525
  3. 3Claude Opus 5 (max)1523
  4. 4Claude Fable 5.1 (max)1522
  5. 5Gemini 3.7 Flash (high)1522
  6. 6Claude Opus 4.6 (high)1516
  7. 7GLM-5.3-Flash1513
  8. 8Claude Opus 4.61507
  9. 9Gemini 3.6 Flash (high)1504
  10. 10GLM-5.3 (max)1504
  11. 11Claude Opus 4.7 (high)1503
  12. 12Kimi K3 (max)1501
  13. 13GPT-5.51500
  14. 14Gemini 3.5 Flash (high)1499
  15. 15Qwen3 8 (max)1498
351 of 351
#
1Claude Fable 5Anthropic152668.32026-09-13
2Claude Opus 5highAnthropic152570.22026-09-13
3Claude Opus 5maxAnthropic152369.92026-09-13
4Claude Fable 5.1maxAnthropic152269.72026-09-13
5Gemini 3.7 FlashhighGoogle152264.62026-09-13
6Claude Opus 4.6highAnthropic151661.32026-09-13
7GLM-5.3-FlashZhipu AIopen ↗151363.92026-09-13
8Claude Opus 4.6Anthropic150763.72026-09-13
9Gemini 3.6 FlashhighGoogle150461.72026-09-13
10GLM-5.3maxZhipu AIopen ↗150465.82026-09-13
11Claude Opus 4.7highAnthropic150362.42026-09-13
12Kimi K3maxMoonshot AIopen ↗150167.22026-09-13
13GPT-5.5OpenAI150063.22026-09-13
14Gemini 3.5 FlashhighGoogle149963.62026-09-13
15Qwen3 8maxAlibaba149866.52026-09-13
16Muse Spark 1.3maxMeta149669.32026-09-13
17GPT-5.4highOpenAI149459.02026-09-13
18Claude Opus 4.8highAnthropic149462.22026-09-13
19Claude Opus 4.7Anthropic149164.52026-09-13
20GPT-5.5highOpenAI149167.02026-09-13
21Gemini 3.1 ProGoogle148963.92026-09-13
22Qwen3 7maxAlibaba148961.02026-09-13
23GPT-5.6 SolxhighOpenAI148868.22026-09-13
24Muse Spark 1.1Meta148665.22026-09-13
25GLM-5.2maxZhipu AIopen ↗148063.92026-09-13
26Kimi K2.6Moonshot AIopen ↗147860.52026-09-13
27Gemini 3.5 FlashmediumGoogle147864.22026-09-13
28Gemini 3 ProGoogle147761.12026-09-13
29GPT-5.6 LunaxhighOpenAI147761.12026-09-13
30Grok 4.5xAI147760.32026-09-13
31MiMo-V2.5-ProXiaomiopen ↗147659.22026-09-13
32GLM-5.1Zhipu AIopen ↗147656.92026-09-13
33Ernie 5.1Baidu1476–2026-09-13
34Claude Sonnet 5highAnthropic147662.22026-09-13
35GPT-5.6 TerraxhighOpenAI147664.32026-09-13
36Gemini 3 FlashGoogle147657.72026-09-13
37Qwen3 6maxAlibaba147563.32026-09-13
38Claude Opus 4.8Anthropic147561.92026-09-13
39Gemma 4 31BGoogleopen ↗147352.72026-09-13
40Kimi K2.5thinkingMoonshot AIopen147058.52026-09-13
41Claude Opus 4.5highAnthropic147057.72026-09-13
42Qwen3 5maxAlibaba1470–2026-09-13
43Deepseek v4 ProhighDeepSeek146960.82026-09-13
44Gemma 4 26B A4BGoogle146857.12026-09-13
45Grok 4.20thinkingxAI146759.92026-09-13
46Qwen3.7 PlusAlibaba146659.92026-09-13
47Qwen3.8 27BAlibabaopen ↗146656.52026-09-13
48Claude Opus 4.5Anthropic146558.22026-09-13
49GPT-5.5 InstantOpenAI146359.22026-09-13
50Claude Sonnet 4.6Anthropic146359.02026-09-13
51Muse SparkMeta146165.72026-09-13
52GPT-5.4OpenAI146159.22026-09-13
53GPT-5.2highOpenAI145855.72026-09-13
54GPT-5.1highOpenAI145658.72026-09-13
55Qwen3.6 PlusAlibabaopen145558.52026-09-13
56Claude Sonnet 4.5highAnthropic145458.82026-09-13
57Muse GlimmerMetaopen ↗145455.72026-09-13
58Gemini 3 FlashminimalGoogle145454.92026-09-13
59GPT-5.2OpenAI145358.42026-09-13
60Grok 4.20 Multi-AgentxAI145359.22026-09-13

Cite as: BenchLeader, “LMArena Maths leaderboard”, https://www.benchleader.com/benchmarks/lmarena_text_math, data as of 19 Sept 2026.

LMArena Maths: questions

What does LMArena Maths measure?
Pairwise votes on prompts classified as mathematics. Scores are reported in Elo-style ratings from pairwise comparisons; higher is better.
Which AI model leads LMArena Maths?
Claude Fable 5 leads LMArena Maths with 1526 as of 19 Sept 2026, ahead of Claude Opus 5 at 1525.
How many models have LMArena Maths results?
351 model configurations have a LMArena Maths result on BenchLeader, all taken from LMArena.
Who runs LMArena Maths and how often is it updated?
LMArena Maths is published by LMArena. BenchLeader re-reads the published results every morning and records the date each result was published.
Does LMArena Maths count toward the BenchLeader Index?
Yes. LMArena Maths contributes to the maths category of the BenchLeader Index, normalised so that 50 is the average of the evaluated models.