All rankings / Coding and agents
Web development
Human preference ratings for the overall category of this Arena task. Scores from different boards are not interchangeable.
150 entries · Arena score · Higher is better
Source publication: 2026-10-28. Retrieved 2026-11-05.
Source: Arena (lmarena-ai) · CC-BY-4.0.
Ranks use confidence bounds. Several models can share a rank; the rows within a shared rank are displayed by score.
| Significance rank | Model / evaluated variant | Arena score | 95% confidence interval | Votes |
|---|---|---|---|---|
| 1 | Claude Opus 5.5 (Max)anthropic | 1,809.84 | 1,797 – 1,822.67 | 3,473 |
| 2 | GPT-6 Astra (Max)openai | 1,773.35 | 1,763.83 – 1,782.88 | 7,210 |
| 2 | Claude Sonnet 5.5 (xHigh)anthropic | 1,769.73 | 1,758.45 – 1,781.01 | 4,469 |
| 2 | Gemini 4 Carbon (High)google | 1,768.15 | 1,753.15 – 1,783.15 | 2,230 |
| 3 | GPT-6.1 Sol (Max)openai | 1,750.57 | 1,739.14 – 1,762 | 4,092 |
| 4 | Claude Fable 5.1 (Max)anthropic | 1,745.85 | 1,737.3 – 1,754.41 | 8,614 |
| 5 | Muse Wonder Full (Max)meta | 1,725.32 | 1,711.46 – 1,739.17 | 2,248 |
| 7 | Claude Sonnet 5.5 (High)anthropic | 1,710.59 | 1,701.39 – 1,719.78 | 6,371 |
| 8 | Kimi K3.1 Pro (Max)moonshot | 1,694.58 | 1,681.62 – 1,707.54 | 2,671 |
| 9 | Claude Opus 5 (Max)anthropic | 1,690.52 | 1,684.42 – 1,696.63 | 19,641 |
| 9 | Gemini 4 Argon (High)google | 1,679.21 | 1,667.78 – 1,690.64 | 3,437 |
| 9 | GPT-6 Sol (Max)openai | 1,678.19 | 1,668.8 – 1,687.58 | 5,714 |
| 10 | Qwen3.8-Max (0902)alibaba | 1,675.94 | 1,669.26 – 1,682.61 | 17,936 |
| 9 | GPT-6.1 Terra (Max)openai | 1,672.65 | 1,657.33 – 1,687.97 | 1,821 |
| 10 | Qwen3.8-Maxalibaba | 1,671.89 | 1,659.82 – 1,683.95 | 3,406 |
| 11 | DeepSeek-V4.1-Flash-20261017 (Max)deepseek | 1,665.04 | 1,657.03 – 1,673.05 | 7,842 |
| 14 | Claude Opus 5 (High)anthropic | 1,658.19 | 1,652.92 – 1,663.46 | 26,887 |
| 14 | Kimi K3 (Max)moonshot | 1,656.39 | 1,650.56 – 1,662.23 | 21,281 |
| 14 | Muse Spark 1.3 (Max)meta | 1,656.23 | 1,648.78 – 1,663.69 | 10,796 |
| 14 | Muse Wonder (Max)meta | 1,652.09 | 1,641.02 – 1,663.17 | 3,724 |
| 20 | Grok 4.7 (xHigh)xai | 1,637.46 | 1,628.7 – 1,646.21 | 6,572 |
| 20 | MiMo-V2.6-Flashxiaomi | 1,637.17 | 1,628.73 – 1,645.62 | 7,068 |
| 21 | Hy4 previewtencent | 1,631.11 | 1,621.46 – 1,640.76 | 4,765 |
| 21 | Qwen3.8-Flash-Nextalibaba | 1,628.32 | 1,620.72 – 1,635.93 | 9,703 |
| 21 | MiMo-V2.6-Proxiaomi | 1,625.42 | 1,617.01 – 1,633.82 | 6,820 |
| 21 | Claude Fable 5 (High)anthropic | 1,624.08 | 1,618.11 – 1,630.06 | 16,951 |
| 21 | GLM-5.3 (Max)zai | 1,622.81 | 1,615.56 – 1,630.06 | 11,150 |
| 23 | Muse Spark 1.3 (xHigh)meta | 1,620.06 | 1,612.67 – 1,627.46 | 10,243 |
| 21 | DeepSeek-V4.1-Flash-20260910 (Max)deepseek | 1,619.16 | 1,608.2 – 1,630.12 | 3,510 |
| 23 | Grok 4.6 (High)xai | 1,616.39 | 1,609.39 – 1,623.38 | 11,966 |
| 23 | GPT-5.6 Sol (xHigh, Codex Harness)openai | 1,615.78 | 1,610.03 – 1,621.54 | 18,884 |
| 28 | GLM-5.3-Flashzai | 1,606.65 | 1,600.31 – 1,612.98 | 21,587 |
| 29 | GLM-5.2 (Max)zai | 1,605.64 | 1,599.74 – 1,611.53 | 17,817 |
| 34 | Qwen3.8-27Balibaba | 1,589.78 | 1,583.52 – 1,596.05 | 18,752 |
| 34 | Gemini 3.7 Flash (High)google | 1,589.67 | 1,583.18 – 1,596.17 | 14,473 |
| 34 | Claude Haiku 5.5 (High)anthropic | 1,588.83 | 1,579.16 – 1,598.5 | 4,758 |
| 34 | Gemini 3.8 Flash (High)google | 1,583.3 | 1,576.27 – 1,590.33 | 12,342 |
| 34 | MiniMax-M3.1-Flashminimax | 1,582.46 | 1,570.93 – 1,593.99 | 3,062 |
| 34 | DeepSeek-V4-Pro (High, 20260813)deepseek | 1,581.86 | 1,572.39 – 1,591.33 | 4,676 |
| 34 | DeepSeek-V4-Flash (High)deepseek | 1,581.84 | 1,572.54 – 1,591.13 | 4,927 |
| 34 | GPT-6 Luna (Max)openai | 1,579.4 | 1,573.6 – 1,585.21 | 27,808 |
| 36 | Step 5 Preview (High)stepfun | 1,571.04 | 1,562.1 – 1,579.98 | 5,966 |
| 42 | Claude Opus 4.7 (High)anthropic | 1,557.54 | 1,551.79 – 1,563.29 | 17,859 |
| 43 | Claude Opus 4.8 (High)anthropic | 1,555.83 | 1,550.07 – 1,561.59 | 18,408 |
| 43 | Claude Opus 4.7anthropic | 1,555.43 | 1,549.81 – 1,561.05 | 18,167 |
| 43 | Grok 4.5xai | 1,553.03 | 1,546.35 – 1,559.71 | 11,018 |
| 44 | Claude Opus 4.6 (High)anthropic | 1,546.24 | 1,540.76 – 1,551.71 | 19,316 |
| 43 | Mistral Large 4mistral | 1,544.31 | 1,533.76 – 1,554.86 | 4,068 |
| 44 | Muse Spark 1.1meta | 1,542.48 | 1,534.85 – 1,550.11 | 7,935 |
| 47 | Claude Sonnet 5 (High)anthropic | 1,540.09 | 1,534.27 – 1,545.91 | 16,832 |
| 47 | Gemini 3.6 Flash (High)google | 1,537.55 | 1,530.05 – 1,545.04 | 8,722 |
| 47 | Claude Opus 4.6anthropic | 1,537.01 | 1,531.65 – 1,542.36 | 20,576 |
| 47 | Claude Opus 4.8anthropic | 1,535.98 | 1,530.28 – 1,541.68 | 17,323 |
| 46 | Muse Spark 1.2 (xHigh)meta | 1,533.67 | 1,520 – 1,547.33 | 2,155 |
| 47 | Muse Spark Open (Max)meta | 1,526.24 | 1,510.94 – 1,541.53 | 1,721 |
| 54 | Claude Sonnet 4.6anthropic | 1,522.65 | 1,517.5 – 1,527.81 | 22,733 |
| 54 | GPT-5.6 Terra (xHigh, Codex Harness)openai | 1,520.13 | 1,514.03 – 1,526.23 | 15,388 |
| 54 | Seed2.1 Pro Previewbytedance | 1,519.1 | 1,513.3 – 1,524.91 | 16,294 |
| 54 | GPT-5.6 Luna (xHigh, Codex Harness)openai | 1,518.91 | 1,512.49 – 1,525.33 | 12,522 |
| 54 | Qwen3.7-Max (20260517)alibaba | 1,515.14 | 1,507.76 – 1,522.51 | 8,855 |
| 55 | GPT-5.5 (xHigh, Codex Harness)openai | 1,512.1 | 1,506.13 – 1,518.07 | 15,553 |
| 56 | Kimi K2.6moonshot | 1,509.22 | 1,501.98 – 1,516.47 | 10,119 |
| 56 | GLM-5.1zai | 1,507.85 | 1,501.07 – 1,514.64 | 10,647 |
| 57 | Hy3tencent | 1,505.43 | 1,497.33 – 1,513.53 | 6,844 |
| 61 | Gemini 3.5 Flash (High)google | 1,499.27 | 1,492.01 – 1,506.54 | 9,331 |
| 63 | Claude Opus 4.5 (High, 32K, 20251101)anthropic | 1,493.66 | 1,485.37 – 1,501.95 | 10,981 |
| 65 | Gemini 3.5 Flash (Medium)google | 1,490.63 | 1,483.97 – 1,497.3 | 12,075 |
| 65 | GPT-5.5 (High, Codex Harness)openai | 1,486.73 | 1,480.81 – 1,492.65 | 17,853 |
| 66 | MiniMax-M3minimax | 1,483.31 | 1,477.9 – 1,488.71 | 21,032 |
| 64 | Spark-X2.6synlan | 1,482 | 1,463.1 – 1,500.9 | 1,210 |
| 65 | Qwen3.6-Max-Previewalibaba | 1,481.82 | 1,469.3 – 1,494.34 | 2,700 |
| 66 | MiMo-V2.5-Proxiaomi | 1,480.44 | 1,475.22 – 1,485.65 | 22,604 |
| 68 | Kimi K2.7 Codemoonshot | 1,473.54 | 1,464.12 – 1,482.95 | 4,950 |
| 70 | Claude Opus 4.5 (20251101)anthropic | 1,469.22 | 1,462.01 – 1,476.44 | 13,362 |
| 71 | DeepSeek-V4-Pro Preview (High)deepseek | 1,464.19 | 1,457.87 – 1,470.51 | 13,087 |
| 68 | GPT-5.4 (High, Codex Harness)openai | 1,463.48 | 1,444.6 – 1,482.35 | 1,345 |
| 72 | Qwen3.6-Plusalibaba | 1,461.63 | 1,456.04 – 1,467.21 | 18,511 |
| 75 | GPT-5.5 (Codex Harness)openai | 1,456.08 | 1,450.19 – 1,461.97 | 15,811 |
| 77 | DeepSeek-V4-Prodeepseek | 1,447.17 | 1,440.83 – 1,453.51 | 13,931 |
| 77 | Gemini 3.1 Pro Previewgoogle | 1,446.65 | 1,441.59 – 1,451.71 | 24,129 |
| 75 | GPT-5.4 (Medium, Codex Harness)openai | 1,443.61 | 1,425.33 – 1,461.9 | 1,374 |
| 78 | Gemini 3 Flashgoogle | 1,440.12 | 1,431.51 – 1,448.74 | 10,845 |
| 78 | Gemini 3 Progoogle | 1,439.97 | 1,431.52 – 1,448.43 | 14,054 |
| 79 | MiMo-V2.5xiaomi | 1,438.27 | 1,432.24 – 1,444.29 | 14,837 |
| 80 | Kimi K2.5 (Thinking)moonshot | 1,436.25 | 1,431.03 – 1,441.46 | 20,008 |
| 78 | Beam 1.0 501B A23Breflection | 1,436.05 | 1,423.27 – 1,448.82 | 2,932 |
| 79 | GLM-5zai | 1,435.57 | 1,427.41 – 1,443.73 | 7,465 |
| 78 | GLM-4.7zai | 1,435.19 | 1,423.11 – 1,447.26 | 3,948 |
| 79 | MiMo-V2-Proxiaomi | 1,434.15 | 1,426 – 1,442.3 | 7,225 |
| 71 | Gemini 3.5 Flash-Litegoogle | 1,434.08 | 1,393.07 – 1,475.1 | 236 |
| 81 | DeepSeek-V4-Flash Preview (High)deepseek | 1,430.48 | 1,423.52 – 1,437.45 | 10,074 |
| 81 | GPT-5 (Medium)openai | 1,417.28 | 1,401.68 – 1,432.87 | 3,089 |
| 81 | GPT-5.2openai | 1,416.42 | 1,393.83 – 1,439.02 | 1,082 |
| 91 | Inklingthinky | 1,415.19 | 1,409.19 – 1,421.19 | 17,073 |
| 91 | Inkling-Smallthinky | 1,409.14 | 1,400.03 – 1,418.24 | 5,295 |
| 91 | Kimi K2.5 (Instant)moonshot | 1,405.57 | 1,393.49 – 1,417.64 | 3,246 |
| 91 | GPT-5.4openai | 1,401.09 | 1,391.13 – 1,411.04 | 4,499 |
| 91 | GLM-5V-Turbozai | 1,400.99 | 1,387.74 – 1,414.23 | 2,394 |
| 92 | Qwen3.5-397B-A17Balibaba | 1,400.22 | 1,395.09 – 1,405.35 | 21,096 |
| 91 | NVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4)nvidia | 1,399.18 | 1,386.28 – 1,412.09 | 3,193 |
| 92 | MiniMax-M2.7minimax | 1,398.19 | 1,392.4 – 1,403.98 | 15,890 |
| 92 | GPT-5.4 mini (High)openai | 1,397.3 | 1,390.76 – 1,403.84 | 11,893 |
| 92 | GPT-5.1 (Medium)openai | 1,395.62 | 1,384.08 – 1,407.16 | 4,891 |
| 93 | Claude Sonnet 4.5 (High, 32K, 20250929)anthropic | 1,393.54 | 1,385.82 – 1,401.26 | 13,213 |
| 93 | Claude Opus 4.1 (20250805)anthropic | 1,390.16 | 1,379.41 – 1,400.9 | 6,904 |
| 95 | MiniMax-M2.5minimax | 1,386.16 | 1,377.82 – 1,394.51 | 7,594 |
| 99 | Claude Sonnet 4.5 (20250929)anthropic | 1,384.91 | 1,377.63 – 1,392.2 | 15,679 |
| 101 | Gemini 3 Flash (Minimal)google | 1,384.49 | 1,379.27 – 1,389.72 | 22,080 |
| 96 | MiniMax-M2.1 Previewminimax | 1,383.98 | 1,374.42 – 1,393.54 | 7,573 |
| 105 | Grok 4.20 Beta (Reasoning, 0309)xai | 1,375.46 | 1,369.38 – 1,381.55 | 15,404 |
| 105 | Solar Pro 4upstage | 1,370.18 | 1,359.82 – 1,380.54 | 4,630 |
| 109 | Gemma 4 31Bgoogle | 1,368.66 | 1,362.53 – 1,374.79 | 15,489 |
| 110 | DeepSeek-V3.2 (Thinking)deepseek | 1,362.52 | 1,353.41 – 1,371.64 | 6,860 |
| 111 | Qwen3.5-122B-A10Balibaba | 1,360.38 | 1,352.72 – 1,368.05 | 8,210 |
| 109 | Gemma 4 26B A4Bgoogle | 1,358.78 | 1,341.83 – 1,375.72 | 1,494 |
| 111 | Qwen3.5-27Balibaba | 1,358.31 | 1,350.26 – 1,366.36 | 7,767 |
| 111 | Grok 4.3xai | 1,357.57 | 1,351.33 – 1,363.81 | 14,906 |
| 110 | Hy3 previewtencent | 1,356.92 | 1,339.55 – 1,374.28 | 1,444 |
| 110 | Muse Glimmermeta | 1,355.06 | 1,339.42 – 1,370.71 | 1,632 |
| 113 | Laguna M.1poolside | 1,349.18 | 1,339.88 – 1,358.48 | 5,359 |
| 117 | GPT-5.1openai | 1,341.34 | 1,332.86 – 1,349.83 | 10,675 |
| 115 | GLM-4.6zai | 1,341.15 | 1,330.34 – 1,351.97 | 6,755 |
| 117 | GPT-5.2-Codexopenai | 1,338.67 | 1,329.73 – 1,347.6 | 6,705 |
| 117 | GPT-5.1-Codexopenai | 1,337.12 | 1,325.54 – 1,348.7 | 4,981 |
| 118 | MiMo-V2-Flash (Non-Thinking)xiaomi | 1,330.62 | 1,320.86 – 1,340.39 | 5,645 |
| 121 | Claude Haiku 4.5 (20251001)anthropic | 1,330.37 | 1,325.44 – 1,335.3 | 28,961 |
| 121 | DeepSeek-V3.2deepseek | 1,325.75 | 1,317.77 – 1,333.73 | 9,327 |
| 124 | Kimi K2 Thinking Turbomoonshot | 1,321.92 | 1,314.45 – 1,329.38 | 13,045 |
| 129 | Laguna XS.2poolside | 1,302.43 | 1,291.69 – 1,313.18 | 4,204 |
| 129 | MiniMax-M2minimax | 1,297.11 | 1,286.17 – 1,308.05 | 6,683 |
| 129 | MiMo-V2-Flash (Thinking)xiaomi | 1,291.7 | 1,275.21 – 1,308.18 | 1,681 |
| 131 | Qwen3-Coder-480B-A35B-Instructalibaba | 1,274.75 | 1,266.96 – 1,282.55 | 12,675 |
| 131 | DeepSeek-V3.2-Expdeepseek | 1,271.44 | 1,257.93 – 1,284.95 | 3,968 |
| 131 | Mistral Medium 3.5mistral | 1,264.01 | 1,249.2 – 1,278.82 | 2,299 |
| 133 | Gemini 3.1 Flash-Lite Previewgoogle | 1,256.26 | 1,249.43 – 1,263.1 | 14,350 |
| 131 | KAT-Coder-Pro V1Organization not supplied | 1,255.83 | 1,236.33 – 1,275.33 | 1,503 |
| 132 | Qwen3.5-35B-A3Balibaba | 1,254.16 | 1,236.7 – 1,271.62 | 1,612 |
| 133 | Qwen3.5-Flashalibaba | 1,243.69 | 1,224.03 – 1,263.36 | 1,321 |
| 133 | GPT-5.1-Codex-Miniopenai | 1,243.55 | 1,221.5 – 1,265.59 | 1,154 |
| 134 | Grok 4.1 Fast (Reasoning)xai | 1,242.08 | 1,230.97 – 1,253.19 | 5,473 |
| 133 | Trinity Large ThinkingOrganization not supplied | 1,238.03 | 1,218.01 – 1,258.05 | 1,352 |
| 133 | Mistral Large 3mistral | 1,232.6 | 1,207.14 – 1,258.06 | 838 |
| 136 | Gemini 2.5 Progoogle | 1,226.1 | 1,210.21 – 1,241.98 | 2,724 |
| 136 | Grok 4.1 (Thinking)xai | 1,214.47 | 1,189.59 – 1,239.35 | 961 |
| 142 | Devstral 2mistral | 1,197.13 | 1,176.39 – 1,217.86 | 1,182 |
| 142 | Granite 4.1 8Bibm | 1,192.48 | 1,174.01 – 1,210.95 | 1,766 |
| 144 | Mercury 2inception-ai | 1,170.38 | 1,145.84 – 1,194.91 | 954 |
| 144 | Grok Code Fast 1xai | 1,166.35 | 1,138.08 – 1,194.61 | 789 |
| 145 | Grok 4 Fast (Reasoning)xai | 1,159.11 | 1,131.69 – 1,186.53 | 742 |
| 150 | Devstral Medium (2507)mistral | 1,071.38 | 1,040.77 – 1,101.99 | 842 |
What this ranking can tell you
- The population is the models in this source board, including variants and models without an available API price.
- Ranks use the published confidence bounds: one plus the number of models whose lower bound exceeds this model’s upper bound. Shared ranks are not proof of equal performance.
Unresolved model identities
- GPT-5.3-Codex (Codex Harness) · 2 source entries. The source publishes different results under this same name without a distinct variant identifier. All ambiguous entries are omitted.
Check the source
The original evidence is retained byte for byte. The downloadable board includes the source URLs, revision and SHA-256 hashes.
Download this board and provenance →