LMArena Math arena · published July 20, 2026
Anthropic leads the LMArena math arena.
As of July 20, 2026, Anthropic leads the LMArena math arena with claude-fable-5 at 1550.2 across 487 battles, according to LMArena.
Current through July 20, 2026. 145 published snapshots. 373 of 376 models resolved (3 held for exact-match review).
| Rank | LMArena rating | Organization | Model | Reported interval | Gap to leader | Battles |
|---|---|---|---|---|---|---|
| 01 | 1550.2 | Anthropic | Claude Fable 5 | 1524.0 to 1576.4 | Leader | 487 |
| 02 | 1518.8 | Anthropic | claude-opus-4-6-thinking | 1507.8 to 1529.9 | -31.4 | 3,231 |
| 03 | 1517.6 | gemini-3.5-flash-high | 1492.2 to 1543.0 | -32.6 | 572 | |
| 04 | 1504.5 | Anthropic | claude-opus-4-7-thinking | 1492.2 to 1516.8 | -45.7 | 2,532 |
| 05 | 1503.5 | xAI | grok-4.5 | 1469.5 to 1537.6 | -46.7 | 287 |
| 06 | 1502.1 | Anthropic | claude-opus-4-6 | 1491.6 to 1512.5 | -48.1 | 3,651 |
| 07 | 1496.7 | OpenAI | gpt-5.5 | 1483.9 to 1509.5 | -53.5 | 2,376 |
| 08 | 1496.3 | Anthropic | claude-opus-4-8-thinking | 1480.8 to 1511.8 | -53.9 | 1,471 |
| 09 | 1494.9 | OpenAI | gpt-5.4-high | 1483.6 to 1506.1 | -55.3 | 3,042 |
| 10 | 1493.4 | Anthropic | Claude Opus 4.7 | 1481.2 to 1505.6 | -56.8 | 2,627 |
| 11 | 1491.3 | Alibaba | qwen3.7-max-preview | 1451.3 to 1531.4 | -58.9 | 219 |
| 12 | 1490.9 | gemini-3.1-pro-preview | 1481.3 to 1500.4 | -59.3 | 4,502 | |
| 13 | 1489.9 | OpenAI | gpt-5.5-high | 1477.4 to 1502.5 | -60.3 | 2,328 |
| 14 | 1481.1 | Meta | muse-spark-1.1 | 1448.5 to 1513.7 | -69.1 | 320 |
| 15 | 1480.5 | Z.ai | glm-5.1 | 1465.2 to 1495.9 | -69.7 | 1,559 |
| 16 | 1479.8 | Moonshot AI | kimi-k2.6 | 1466.2 to 1493.4 | -70.4 | 1,931 |
| 17 | 1479.5 | gemini-3.5-flash-medium | 1456.6 to 1502.5 | -70.7 | 621 | |
| 18 | 1478.3 | Baidu | ernie-5.1 | 1464.5 to 1492.0 | -71.9 | 1,899 |
| 19 | 1478.0 | Anthropic | claude-opus-4-8 | 1462.6 to 1493.5 | -72.2 | 1,487 |
| 20 | 1477.7 | Gemini 3 Pro | 1466.2 to 1489.1 | -72.5 | 2,645 | |
| 21 | 1476.8 | Alibaba | qwen3.6-max-preview | 1446.8 to 1506.7 | -73.4 | 359 |
| 22 | 1476.3 | gemini-3-flash | 1463.2 to 1489.5 | -73.9 | 1,991 | |
| 23 | 1475.3 | Xiaomi | mimo-v2.5-pro | 1461.8 to 1488.8 | -74.9 | 2,036 |
| 24 | 1475.0 | Z.ai | glm-5.2 (max) | 1454.9 to 1495.2 | -75.2 | 829 |
| 25 | 1474.0 | Alibaba | qwen3.7-plus | 1455.3 to 1492.7 | -76.2 | 996 |
Frontier race
Category Kings: Math arena frontier by organization
Anthropic holds the math arena frontier at 1550 as of July 20, 2026.
Method: Frontier is the maximum resolved LMArena rating per organization on each publish date. Organizations limited to the union of the seven tracked homepage providers and organizations in the latest board top 25 by officialRank. Quarantined dates are excluded, and this category series keeps its own method epoch and is never joined to the legacy archive.
| Organization | Latest model | Latest LMArena rating | First tracked date |
|---|---|---|---|
| Anthropic | anthropic/claude-fable-5 | 1550.2 | January 5, 2025 |
| google/gemini-3.5-flash-high | 1517.6 | January 5, 2025 | |
| xAI | xai/grok-4.5 | 1503.5 | January 5, 2025 |
| OpenAI | openai/gpt-5.5 | 1496.7 | January 5, 2025 |
| Alibaba | alibaba/qwen3.7-max-preview | 1491.3 | January 5, 2025 |
| Meta | meta/muse-spark-1.1 | 1481.1 | January 5, 2025 |
| Z.ai | zai/glm-5.1 | 1480.5 | January 5, 2025 |
| Moonshot AI | moonshot/kimi-k2.6 | 1479.8 | July 17, 2025 |
| Baidu | baidu/ernie-5.1 | 1478.3 | November 16, 2025 |
| Xiaomi | xiaomi/mimo-v2.5-pro | 1475.3 | December 30, 2025 |
| DeepSeek | deepseek/deepseek-v4-pro-thinking | 1469.8 | January 5, 2025 |
| Mistral AI | mistral/mistral-medium-3.5 | 1432.6 | January 5, 2025 |
Definitions
What these terms mean.
- Math arena
- The LMArena style-controlled text battles filtered to prompts categorized as math. It reflects human preference on math prompts, not verified proof correctness.
- Frontier by organization
- The highest resolved score for each organization on each publish date. It tracks how far each lab has pushed this category, not an average of all its models.
- Confidence interval
- The reported 95% range LMArena publishes with each rating. Overlapping intervals mean this snapshot does not separate those ranks decisively.
Follow the evidence
Compare this board against the overall LMArena rankings, shortlist candidates with the Reasoning and research guide,read the Anthropic lab file, and see how rankings are selected. Model profiles on this board: Claude Fable 5, Claude Opus 4.7, Gemini 3 Pro.
Math arena questions
Which model leads the Math arena right now?
According to the LMArena math board published July 20, 2026, Anthropic leads with claude-fable-5 at 1550.2, across 487 battles. That is 31.4 points ahead of the next board entry inside this one snapshot.
Is this a Model Gauntlet math ranking?
No. This board mirrors one named source, the LMArena math arena, published July 20, 2026. Model Gauntlet publishes no composite or consensus score and runs no evaluation of its own.
How many models and organizations are in the math board?
The reviewed artifact covers 376 models across 60 organizations, with 373 models resolved to exact identities and 3 held for review. The history spans 145 snapshots from January 5, 2025 to July 20, 2026.
Why do some scores have overlapping confidence intervals?
LMArena publishes a 95% confidence interval with each rating. Where two intervals overlap, this single snapshot does not separate those ranks decisively, so treat small math gaps with caution.