Design Arena: agents building full-stack apps leaderboard
- GPT-6 Astra leads
- 1357
- models measured
- 47
- lead over #2
- +31 Elo
- median
- 1181
Leader on Design Arena: agents building full-stack apps · read October 21
On October 21, 2026, 47 of our catalog's models had a Design Arena: agents building full-stack apps result, and OpenAI GPT-6 Astra led with 1357.
Design Arena shows people two designs made by two models from the same prompt (a full-stack app, built by the model as an agent) and asks which is better. The number is the Elo rating built from those votes, as Design Arena publishes it.
Role in the index
We collect this benchmark and show it as supporting evidence. It is not part of the Vibe Coding Index, because it measures something other than coding and the index is about coding. Read the methodology
| # | Model | Result | Battles | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|---|
| 1 | GPT-6 AstraOpenAI | 1357 | 553 | 88.7 | $50.00 | |
| 2 | Claude Fable 5.1Anthropic | 1326 | 509 | 89.1 | $50.00 | |
| 3 | Kimi K3Moonshot AI | 1315 | 1,316 | 61.8 | $11.936 | |
| 4 | Claude Opus 5Anthropic | 1306 | 1,195 | 81.8 | $25.00 | |
| 5 | Qwen3.8 Max (0902)Qwen | 1305 | 1,327 | 63.5 | $6.00 | |
| 6 | Muse Spark 1.3Meta | 1295 | 4,009 | 79.7 | $4.25 | |
| 7 | Claude Fable 5Anthropic | 1280 | 2,118 | 82.1 | $50.00 | |
| 8 | Grok 4.6xAI | 1264 | Not published | 76.4 | $6.00 | |
| 9 | GPT-5.6 SolOpenAI | 1261 | 1,977 | 76.7 | $20.00 | |
| 10 | Claude Opus 4.8Anthropic | 1249 | 2,700 | 66.0 | $25.00 | |
| 11 | Claude Sonnet 5Anthropic | 1249 | 3,747 | 63.5 | $10.00 | |
| 12 | Grok 4.5xAI | 1249 | Not published | 75.5 | $6.00 | |
| 13 | Gemini 3.8 FlashGoogle | 1244 | 1,155 | 70.5 | $4.125 | |
| 14 | GLM 5.2Z.ai | 1241 | 2,760 | 57.7 | $7.00 | |
| 15 | GPT-6 SolOpenAI | 1238 | 1,172 | 82.8 | $10.00 | |
| 16 | Claude Opus 4.6Anthropic | 1220 | 2,517 | 66.9 | $25.00 | |
| 17 | Claude Sonnet 4.6Anthropic | 1213 | 2,030 | 57.8 | $15.00 | |
| 18 | Muse Spark 1.2Meta | 1210 | 1,907 | 65.1 | $4.25 | |
| 19 | Gemini 3.7 FlashGoogle | 1208 | 682 | 69.9 | $3.75 | |
| 20 | Muse Spark 1.1Meta | 1199 | Not published | 48.9 | $4.25 | |
| 21 | Gemini 3.5 FlashGoogle | 1197 | 4,032 | 42.8 | $9.00 | |
| 22 | MiniMax M3MiniMax | 1197 | 3,674 | 57.8 | $1.20 | |
| 23 | Qwen3.7 MaxQwen | 1191 | 1,417 | 45.1 | $4.425 | |
| 24 | Gemini 3.6 FlashGoogle | 1181 | 2,935 | 48.3 | $3.75 | |
| 25 | Kimi K2.7 CodeMoonshot AI | 1175 | 2,794 | 39.1 | $3.35 | |
| 26 | GLM 5.1Z.ai | 1167 | 3,901 | 44.0 | $4.40 | |
| 27 | GPT-6 LunaOpenAI | 1158 | 1,052 | 71.4 | $0.50 | |
| 28 | Claude Opus 4.5Anthropic | 1156 | 1,396 | 66.8 | $25.00 | |
| 29 | Kimi K2.6Moonshot AI | 1155 | 5,980 | 44.2 | $3.41 | |
| 30 | GLM 5V TurboZ.ai | 1149 | 3,947 | 28.8 | $4.00 | |
| 31 | GLM 5Z.ai | 1123 | 4,765 | 59.7 | $3.20 | |
| 32 | Kimi K2.5Moonshot AI | 1114 | 1,787 | 58.5 | $2.50 | |
| 33 | GPT-5.5OpenAI | 1089 | 3,539 | 67.8 | $30.00 | |
| 34 | GPT-5.1-CodexOpenAI | 1086 | Not published | 50.5 | $10.00 | |
| 35 | Gemini 3.1 Pro PreviewGoogle | 1068 | 4,827 | 35.3 | $12.00 | |
| 36 | Gemini 3 Flash PreviewGoogle | 1063 | 2,359 | 75.8 | $3.00 | |
| 37 | GPT-5.6 TerraOpenAI | 1061 | 2,589 | 69.0 | $12.00 | |
| 38 | Grok 4.20xAI | 1054 | Not published | Not benchmarked | $2.50 | |
| 39 | Claude Sonnet 4.5Anthropic | 1050 | 629 | 59.1 | $15.00 | |
| 40 | GLM 4.7Z.ai | 1050 | 1,639 | 33.6 | $1.98 | |
| 41 | GPT-5.2OpenAI | 1043 | 2,165 | 58.9 | $14.00 | |
| 42 | GLM 4.6Z.ai | 1029 | 1,885 | 52.2 | $2.20 | |
| 43 | GPT-5.4OpenAI | 1017 | 2,152 | 37.7 | $15.00 | |
| 44 | Grok 4.3xAI | 1006 | 3,448 | 22.5 | $2.50 | |
| 45 | GPT-5.2-CodexOpenAI | 993 | 2,158 | 55.1 | $14.00 | |
| 46 | GPT-5.3-CodexOpenAI | 983 | 1,110 | 29.5 | $14.00 | |
| 47 | DeepSeek V4 Pro 0423DeepSeek | 948 | Not published | 35.3 | $1.9105 |
Only models in our catalog are listed, and each appears once with its best published run. 141 rows were read from the source on October 21, 2026, and rows that match no catalog model are left out. Every other column is a figure the board publishes for the same run.