LMArena Coding arena · published July 20, 2026
Anthropic leads the LMArena coding arena.
As of July 20, 2026, Anthropic leads the LMArena coding arena with claude-opus-4-7-thinking at 1553.0 across 14,244 battles, according to LMArena.
Current through July 20, 2026. 165 published snapshots. 380 of 384 models resolved (4 held for exact-match review).
| Rank | LMArena rating | Organization | Model | Reported interval | Gap to leader | Battles |
|---|---|---|---|---|---|---|
| 01 | 1553.0 | Anthropic | claude-opus-4-7-thinking | 1546.4 to 1559.6 | Leader | 14,244 |
| 02 | 1550.1 | Anthropic | claude-opus-4-6-thinking | 1543.9 to 1556.2 | -2.9 | 16,120 |
| 03 | 1548.7 | Anthropic | Claude Opus 4.7 | 1542.2 to 1555.2 | -4.3 | 14,491 |
| 04 | 1547.7 | Anthropic | Claude Fable 5 | 1535.9 to 1559.6 | -5.3 | 2,725 |
| 05 | 1547.6 | Anthropic | claude-opus-4-6 | 1541.8 to 1553.5 | -5.4 | 18,333 |
| 06 | 1535.1 | Anthropic | claude-opus-4-8-thinking | 1527.3 to 1542.9 | -17.9 | 8,366 |
| 07 | 1533.0 | Meta | muse-spark-1.1 | 1519.5 to 1546.5 | -20.0 | 2,087 |
| 08 | 1530.7 | Anthropic | claude-opus-4-8 | 1522.9 to 1538.5 | -22.3 | 8,589 |
| 09 | 1530.3 | Anthropic | claude-opus-4-5-20251101-thinking-32k | 1522.9 to 1537.6 | -22.7 | 7,621 |
| 10 | 1527.4 | Anthropic | claude-sonnet-4-6 | 1521.2 to 1533.6 | -25.6 | 15,590 |
| 11 | 1526.3 | Alibaba | qwen3.7-max-preview | 1507.9 to 1544.8 | -26.7 | 1,123 |
| 12 | 1526.1 | OpenAI | gpt-5.6-sol-xhigh | 1510.5 to 1541.7 | -26.9 | 1,550 |
| 13 | 1525.0 | Meta | Muse Spark | 1514.8 to 1535.2 | -28.0 | 3,777 |
| 14 | 1523.6 | Anthropic | claude-sonnet-5-high | 1513.0 to 1534.1 | -29.4 | 3,543 |
| 15 | 1522.7 | Anthropic | claude-opus-4-5-20251101 | 1517.1 to 1528.2 | -30.3 | 17,316 |
| 16 | 1521.1 | gemini-3.1-pro-preview | 1515.7 to 1526.6 | -31.9 | 22,888 | |
| 17 | 1521.0 | OpenAI | gpt-5.4-high | 1514.8 to 1527.1 | -32.0 | 15,756 |
| 18 | 1520.5 | Z.ai | glm-5.1 | 1513.1 to 1527.9 | -32.5 | 8,422 |
| 19 | 1519.8 | Anthropic | claude-sonnet-4-5-20250929-thinking-32k | 1514.8 to 1524.8 | -33.2 | 19,300 |
| 20 | 1519.6 | xAI | grok-4.5 | 1506.1 to 1533.2 | -33.4 | 2,076 |
| 21 | 1519.3 | OpenAI | gpt-5.5-high | 1512.4 to 1526.1 | -33.7 | 12,460 |
| 22 | 1518.8 | Xiaomi | mimo-v2.5-pro | 1511.7 to 1525.9 | -34.2 | 11,355 |
| 23 | 1518.7 | Gemini 3 Pro | 1511.6 to 1525.8 | -34.3 | 8,576 | |
| 24 | 1515.4 | OpenAI | gpt-5.2-chat-latest-20260210 | 1508.4 to 1522.4 | -37.6 | 9,160 |
| 25 | 1514.9 | Moonshot AI | kimi-k2.6 | 1507.7 to 1522.0 | -38.1 | 10,419 |
Frontier race
Category Kings: Coding arena frontier by organization
Anthropic holds the coding arena frontier at 1553 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-opus-4-7-thinking | 1553.0 | September 15, 2024 |
| Meta | meta/muse-spark-1.1 | 1533.0 | September 15, 2024 |
| Alibaba | alibaba/qwen3.7-max-preview | 1526.3 | September 15, 2024 |
| OpenAI | openai/gpt-5.6-sol-xhigh | 1526.1 | September 15, 2024 |
| google/gemini-3.1-pro-preview | 1521.1 | September 15, 2024 | |
| Z.ai | zai/glm-5.1 | 1520.5 | September 15, 2024 |
| xAI | xai/grok-4.5 | 1519.6 | September 15, 2024 |
| Xiaomi | xiaomi/mimo-v2.5-pro | 1518.8 | December 23, 2025 |
| Moonshot AI | moonshot/kimi-k2.6 | 1514.9 | July 17, 2025 |
| DeepSeek | deepseek/deepseek-v4-pro | 1501.4 | September 15, 2024 |
| Mistral AI | mistral/mistral-medium-3.5 | 1479.1 | September 15, 2024 |
Definitions
What these terms mean.
- Coding arena
- The LMArena style-controlled text battles filtered to prompts categorized as coding. Model Gauntlet mirrors this named subset and does not run code evaluations itself.
- 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 Coding shortlist guide,read the Anthropic lab file, and see how rankings are selected. Model profiles on this board: Claude Opus 4.7, Claude Fable 5, Muse Spark.
Other coding boards
Aider polyglot mirrors a different coding subject.
The Aider polyglot leaderboard ranks full model-and-editor-system configurations on code editing, not bare models. A resolved entry names only the model portion of a system configuration, so it is never a bare-model score and is not comparable to this LMArena coding arena. It appears as its own labeled column with attribution.
See the Aider polyglot column →Coding arena questions
Which model leads the Coding arena right now?
According to the LMArena coding board published July 20, 2026, Anthropic leads with claude-opus-4-7-thinking at 1553.0, across 14,244 battles. That is 2.9 points ahead of the next board entry inside this one snapshot.
Is this a Model Gauntlet coding ranking?
No. This board mirrors one named source, the LMArena coding 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 coding board?
The reviewed artifact covers 384 models across 61 organizations, with 380 models resolved to exact identities and 4 held for review. The history spans 165 snapshots from September 15, 2024 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 coding gaps with caution.