llmrun
Coding

LMArena WebDev Leaderboard

Code Arena (formerly WebDev Arena) is Arena's head-to-head coding leaderboard: two models each build a real web app or UI from the same prompt, and people vote blind on which result is better. The rating shown is a relative Bradley-Terry score, not a percentage — compare it only to other models on this board.

Source: lmarena55 open models ranked+91 proprietaryData through Oct 2026

Open modelsAll models

All models ranked on LMArena WebDev

Proprietary / closed models are shown dimmed — you can't run them locally, but they show where the open field stands.

#ModelScoreVRAM @ Q4
1Claude Opus 5.5 (Max) · proprietary
1809.8
API only
2GPT 6 Astra (Max) · proprietary
1773.4
API only
3Claude Sonnet 5.5 (xHigh) · proprietary
1769.7
API only
4Gemini 4 Carbon (High) · proprietary
1768.2
API only
5GPT 6.1 Sol (Max) · proprietary
1750.6
API only
6Claude Fable 5.1 (Max) · proprietary
1745.9
API only
7Muse Wonder Full (Max) · proprietary
1725.3
API only
8Claude Sonnet 5.5 (High) · proprietary
1710.6
API only
9Kimi K3.1 Pro (Max) · proprietary
1694.6
API only
10Claude Opus 5 (Max) · proprietary
1690.5
API only
11Gemini 4 Argon (High) · proprietary
1679.2
API only
12GPT 6 Sol (Max) · proprietary
1678.2
API only
13Qwen3.8 Max (0902) · proprietary
1675.9
API only
14GPT 6.1 Terra (Max) · proprietary
1672.7
API only
15Qwen3.8 Max · proprietary
1671.9
API only
16DeepSeek V4.1 Flash 1017 · 763.2B
1665.0
~458 GB
17Claude Opus 5 (High) · proprietary
1658.2
API only
18Kimi K3 · 2779.9B
1656.4
~1674 GB
19Muse Spark 1.3 (Max) · proprietary
1656.2
API only
20Muse Wonder (Max) · proprietary
1652.1
API only
21Grok 4.7 (xHigh) · proprietary
1637.5
API only
22MiMo V2.6 Flash RL · 309B
1637.2
~186 GB
23Hy4 Preview · 780.0B
1631.1
~469 GB
24Qwen3.8 Flash Next · 180.0B
1628.3
~108 GB
25MiMo V2.6 Pro RL · 1020B
1625.4
~613 GB
26Claude Fable 5 (High) · proprietary
1624.1
API only
27GLM 5.3 · 753.3B
1622.8
~456 GB
28Muse Spark 1.3 (xHigh) · proprietary
1620.1
API only
29DeepSeek V4.1 Flash · 763.2B
1619.2
~458 GB
30Grok 4.6 (High) · proprietary
1616.4
API only
31GPT 5.6 Sol (xHigh, Codex Harness) · proprietary
1615.8
API only
32GLM 5.3 Flash · 321.3B
1606.6
~195 GB
33GLM 5.2 · 753.3B
1605.6
~456 GB
34Qwen3.8 27B · 27.8B
1589.8
~17 GB
35Gemini 3.7 Flash (High) · proprietary
1589.7
API only
36Claude Haiku 5.5 (High) · proprietary
1588.8
API only
37Gemini 3.8 Flash (High) · proprietary
1583.3
API only
38MiniMax M3.1 Flash · 428B
1582.5
~257 GB
39DeepSeek V4 Pro 0813 · 1650.5B
1581.9
~991 GB
40DeepSeek V4 Flash · 290.9B
1581.8
~175 GB
41GPT 6 Luna (Max) · proprietary
1579.4
API only
42Step 5 Preview (High) · proprietary
1571.0
API only
43Claude Opus 4.7 (High) · proprietary
1557.5
API only
44Claude Opus 4.8 (High) · proprietary
1555.8
API only
45Claude Opus 4.7 · proprietary
1555.4
API only
46Grok 4.5 · proprietary
1553.0
API only
47Claude Opus 4.6 (High) · proprietary
1546.2
API only
48Mistral Large 4 · proprietary
1544.3
API only
49Muse Spark 1.1 · proprietary
1542.5
API only
50Claude Sonnet 5 (High) · proprietary
1540.1
API only
51Gemini 3.6 Flash (High) · proprietary
1537.5
API only
52Claude Opus 4.6 · proprietary
1537.0
API only
53Claude Opus 4.8 · proprietary
1536.0
API only
54Muse Spark 1.2 (xHigh) · proprietary
1533.7
API only
55Muse Spark Open 816B A42B · 822.0B
1526.2
~543 GB
56Claude Sonnet 4.6 · proprietary
1522.7
API only
57GPT 5.6 Terra (xHigh, Codex Harness) · proprietary
1520.1
API only
58Seed2.1 Pro Preview · proprietary
1519.1
API only
59GPT 5.6 Luna (xHigh, Codex Harness) · proprietary
1518.9
API only
60Qwen3.7 Max (20260517) · proprietary
1515.1
API only
61GPT 5.5 (xHigh, Codex Harness) · proprietary
1512.1
API only
62Kimi K2.6 · 1026.9B
1509.2
~620 GB
63GLM 5.1 · 753.9B
1507.9
~457 GB
64Hy3 · 298.8B
1505.4
~180 GB
65Gemini 3.5 Flash (High) · proprietary
1499.3
API only
66Claude Opus 4.5 (High, 32K, 20251101) · proprietary
1493.7
API only
67Gemini 3.5 Flash (Medium) · proprietary
1490.6
API only
68GPT 5.5 (High, Codex Harness) · proprietary
1486.7
API only
69MiniMax M3 · 427.0B
1483.3
~257 GB
70Spark X2.6 · proprietary
1482.0
API only
71Qwen3.6 Max Preview · proprietary
1481.8
API only
72MiMo V2.5 Pro · 1023.2B
1480.4
~614 GB
73Kimi K2.7 Code · 1026.9B
1473.5
~620 GB
74Claude Opus 4.5 (20251101) · proprietary
1469.2
API only
75DeepSeek V4 Pro · 1598.8B
1464.2
~960 GB
76GPT 5.4 (High, Codex Harness) · proprietary
1463.5
API only
77Qwen3.6 Plus · proprietary
1461.6
API only
78GPT 5.5 (Codex Harness) · proprietary
1456.1
API only
79Gemini 3.1 Pro Preview · proprietary
1446.6
API only
80GPT 5.4 (Medium, Codex Harness) · proprietary
1443.6
API only
81Gemini 3 Flash · proprietary
1440.1
API only
82Gemini 3 Pro · proprietary
1440.0
API only
83MiMo V2.5 · 310.8B
1438.3
~187 GB
84Kimi K2.5 · 1026.9B
1436.2
~620 GB
85Beam 1.0 501B A23B · 501.3B
1436.0
~301 GB
86GLM 5 · 753.9B
1435.6
~457 GB
87GLM 4.7 · 358.3B
1435.2
~216 GB
88MiMo V2 Pro · proprietary
1434.2
API only
89Gemini 3.5 Flash Lite · proprietary
1434.1
API only
90GPT 5 (Medium) · proprietary
1417.3
API only
91GPT 5.2 · proprietary
1416.4
API only
92Inkling · 952.4B
1415.2
~572 GB
93Inkling Small · 266.0B
1409.1
~160 GB
94GPT 5.3 Codex (Codex Harness) · proprietary
1406.5
API only
95GPT 5.4 · proprietary
1401.1
API only
96GLM 5V Turbo · proprietary
1401.0
API only
97Qwen3.5 397B A17B · 403.4B
1400.2
~242 GB
98NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 · 124.5B
1399.2
~82 GB
99MiniMax M2.7 · 228.7B
1398.2
~138 GB
100GPT 5.4 mini (High) · proprietary
1397.3
API only
101GPT 5.1 (Medium) · proprietary
1395.6
API only
102Claude Sonnet 4.5 (High, 32K, 20250929) · proprietary
1393.5
API only
103Claude Opus 4.1 (20250805) · proprietary
1390.2
API only
104MiniMax M2.5 · 228.7B
1386.2
~138 GB
105Claude Sonnet 4.5 (20250929) · proprietary
1384.9
API only
106Gemini 3 Flash (Minimal) · proprietary
1384.5
API only
107MiniMax M2.1 · 228.7B
1384.0
~138 GB
108Grok 4.20 Beta (Reasoning, 0309) · proprietary
1375.5
API only
109Solar Pro 4 · proprietary
1370.2
API only
110Gemma 4 31B IT · 31.3B
1368.7
~20 GB
111DeepSeek V3.2 · 685.4B
1362.5
~415 GB
112Qwen3.5 122B A10B · 125.1B
1360.4
~75 GB
113Gemma 4 26B A4B IT · 25.8B
1358.8
~16 GB
114Qwen3.5 27B · 27.8B
1358.3
~17 GB
115Grok 4.3 · proprietary
1357.6
API only
116Hy3 Preview · 298.8B
1356.9
~180 GB
117Muse Glimmer 30B · 29.8B
1355.1
~18 GB
118Laguna M.1 · 225.8B
1349.2
~136 GB
119GPT 5.1 · proprietary
1341.3
API only
120GLM 4.6 · 356.8B
1341.2
~215 GB
121GPT 5.2 Codex · proprietary
1338.7
API only
122GPT 5.1 Codex · proprietary
1337.1
API only
123MiMo v2 Flash · 309.8B
1330.6
~186 GB
124Claude Haiku 4.5 (20251001) · proprietary
1330.4
API only
125Kimi K2 Thinking · 1026.4B
1321.9
~620 GB
126Laguna XS.2 · 33.4B
1302.4
~20 GB
127MiniMax M2 · 228.7B
1297.1
~138 GB
128Qwen3 Coder 480B A35B Instruct · 480.2B
1274.8
~289 GB
129DeepSeek V3.2 Exp · 685.4B
1271.4
~415 GB
130Mistral Medium 3.5 128B · 127.7B
1264.0
~78 GB
131Gemini 3.1 Flash Lite Preview · proprietary
1256.3
API only
132KAT Coder Pro V1 · proprietary
1255.8
API only
133Qwen3.5 35B A3B · 36.0B
1254.2
~22 GB
134Qwen3.5 Flash · proprietary
1243.7
API only
135GPT 5.1 Codex Mini · proprietary
1243.5
API only
136Grok 4.1 Fast (Reasoning) · proprietary
1242.1
API only
137Trinity Large Thinking · 398.6B
1238.0
~798 GB
138Mistral Large 3 675B Instruct 2512 · 675B
1232.6
~446 GB
139Gemini 2.5 Pro · proprietary
1226.1
API only
140Grok 4.1 (Thinking) · proprietary
1214.5
API only
141Devstral 2 123B Instruct 2512 · 125.0B
1197.1
~76 GB
142Granite 4.1 8B · 8.8B
1192.5
~5.9 GB
143Mercury 2 · proprietary
1170.4
API only
144Grok Code Fast 1 · proprietary
1166.3
API only
145Grok 4 Fast (Reasoning) · proprietary
1159.1
API only
146Devstral Medium (2507) · proprietary
1071.4
API only

Score vs model size

Which models give the most quality for their size — the ones worth running locally.

10B100B1Tmodel size (log scale) →1665.01192.5Kimi K3 · 2.8T · 1656.4Hy4 Preview · 780B · 1631.1MiMo V2.6 Pro RL · 1T · 1625.4GLM 5.3 · 753B · 1622.8DeepSeek V4.1 Flash · 763B · 1619.2GLM 5.3 Flash · 321B · 1606.6GLM 5.2 · 753B · 1605.6MiniMax M3.1 Flash · 428B · 1582.5DeepSeek V4 Pro 0813 · 1.7T · 1581.9DeepSeek V4 Flash · 291B · 1581.8Muse Spark Open 816B A42B · 822B · 1526.2Kimi K2.6 · 1T · 1509.2GLM 5.1 · 754B · 1507.9Hy3 · 299B · 1505.4MiniMax M3 · 427B · 1483.3MiMo V2.5 Pro · 1T · 1480.4Kimi K2.7 Code · 1T · 1473.5DeepSeek V4 Pro · 1.6T · 1464.2MiMo V2.5 · 311B · 1438.3Kimi K2.5 · 1T · 1436.2Beam 1.0 501B A23B · 501B · 1436.0GLM 5 · 754B · 1435.6GLM 4.7 · 358B · 1435.2Inkling · 952B · 1415.2Inkling Small · 266B · 1409.1Qwen3.5 397B A17B · 403B · 1400.2NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 · 124B · 1399.2MiniMax M2.7 · 229B · 1398.2MiniMax M2.5 · 229B · 1386.2MiniMax M2.1 · 229B · 1384.0Gemma 4 31B IT · 31B · 1368.7DeepSeek V3.2 · 685B · 1362.5Qwen3.5 122B A10B · 125B · 1360.4Qwen3.5 27B · 28B · 1358.3Hy3 Preview · 299B · 1356.9Muse Glimmer 30B · 30B · 1355.1Laguna M.1 · 226B · 1349.2GLM 4.6 · 357B · 1341.2MiMo v2 Flash · 310B · 1330.6Kimi K2 Thinking · 1T · 1321.9Laguna XS.2 · 33B · 1302.4MiniMax M2 · 229B · 1297.1Qwen3 Coder 480B A35B Instruct · 480B · 1274.8DeepSeek V3.2 Exp · 685B · 1271.4Mistral Medium 3.5 128B · 128B · 1264.0Qwen3.5 35B A3B · 36B · 1254.2Trinity Large Thinking · 399B · 1238.0Mistral Large 3 675B Instruct 2512 · 675B · 1232.6Devstral 2 123B Instruct 2512 · 125B · 1197.1Granite 4.1 8B · 9B · 1192.5Granite 4.1 8BGemma 4 26B A4B IT · 26B · 1358.8Gemma 4 26B A4B ITQwen3.8 27B · 28B · 1589.8Qwen3.8 27BQwen3.8 Flash Next · 180B · 1628.3Qwen3.8 Flash NextMiMo V2.6 Flash RL · 309B · 1637.2MiMo V2.6 Flash RLDeepSeek V4.1 Flash 1017 · 763B · 1665.0DeepSeek V4.1 Flash 1…
Each dot is a model. Up = higher score, left = smaller (easier to run locally). The dashed line marks the efficiency frontier — the best score you can get at each size or smaller.
  • Granite 4.1 8B, 9B, score 1192.5 — on the efficiency frontier (best score at its size or smaller).
  • Gemma 4 26B A4B IT, 26B, score 1358.8 — on the efficiency frontier (best score at its size or smaller).
  • Qwen3.8 27B, 28B, score 1589.8 — on the efficiency frontier (best score at its size or smaller).
  • Qwen3.8 Flash Next, 180B, score 1628.3 — on the efficiency frontier (best score at its size or smaller).
  • MiMo V2.6 Flash RL, 309B, score 1637.2 — on the efficiency frontier (best score at its size or smaller).
  • DeepSeek V4.1 Flash 1017, 763B, score 1665.0 — on the efficiency frontier (best score at its size or smaller).

LMArena WebDev: frequently asked questions

What is the best open LLM on LMArena WebDev?
DeepSeek V4.1 Flash 1017 is the top open model on LMArena WebDev, scoring 1665.0. Among all models tested — including proprietary ones — it ranks #16. The top model overall is Claude Opus 5.5 (Max) (Anthropic) at 1809.8.
What's the best LMArena WebDev model you can run on a 24 GB GPU?
Qwen3.8 27B is the highest-scoring open model that fits in 24 GB at 4-bit quantization (~17 GB), scoring 1589.8 on LMArena WebDev.
What's the best LMArena WebDev model you can run on a 12 GB GPU?
Granite 4.1 8B is the highest-scoring open model that fits in 12 GB at 4-bit quantization (~5.9 GB), scoring 1192.5 on LMArena WebDev.
Can open models match proprietary models on LMArena WebDev?
Not quite on LMArena WebDev: the strongest proprietary model (Claude Opus 5.5 (Max)) scores 1809.8, ahead of the best open model (DeepSeek V4.1 Flash 1017) at 1665.0 — but you can run the open one yourself.

Scores aggregated from lmarena. llmrun does not run this benchmark — see the source for methodology, or the about benchmarks for what it measures.