Design Arena: agents building Godot games leaderboard
- GPT-6 Astra leads
- 1348
- models measured
- 37
- lead over #2
- +21 Elo
- median
- 1180
Leader on Design Arena: agents building Godot games · read October 22
On October 22, 2026, 37 of our catalog's models had a Design Arena: agents building Godot games result, and OpenAI GPT-6 Astra led with 1348, ahead of Anthropic Claude Fable 5.1 at 1327.
Design Arena shows people two designs made by two models from the same prompt (a game in the Godot engine, 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
Design Arena: agents building Godot games is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Battles | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|---|
| 1 | GPT-6 AstraOpenAI | 1348 | 328 | 85.1 | $50.00 | |
| 2 | Claude Fable 5.1Anthropic | 1327 | 629 | 84.2 | $50.00 | |
| 3 | Claude Fable 5Anthropic | 1292 | 1,379 | 79.3 | $50.00 | |
| 4 | Claude Opus 5Anthropic | 1279 | 793 | 83.0 | $25.00 | |
| 5 | Grok 4.6xAI | 1276 | Not published | 65.7 | $6.00 | |
| 6 | Gemini 3.8 FlashGoogle | 1273 | 664 | 61.6 | $3.75 | |
| 7 | Gemini 3.7 FlashGoogle | 1264 | 461 | 58.0 | $3.75 | |
| 8 | GPT-5.6 SolOpenAI | 1260 | 595 | 73.5 | $20.00 | |
| 9 | Muse Spark 1.3Meta | 1241 | Not published | 57.4 | $4.25 | |
| 10 | Gemini 3.1 Pro PreviewGoogle | 1236 | Not published | 32.3 | $12.00 | |
| 11 | Qwen3.8 Max (0902)Qwen | 1234 | 453 | 62.4 | $6.00 | |
| 12 | Claude Opus 4.8Anthropic | 1231 | 1,659 | 65.0 | $25.00 | |
| 13 | Grok 4.5xAI | 1229 | Not published | 58.6 | $6.00 | |
| 14 | Claude Sonnet 4.6Anthropic | 1226 | Not published | 57.3 | $15.00 | |
| 15 | Claude Sonnet 5Anthropic | 1225 | 1,573 | 57.7 | $10.00 | |
| 16 | Kimi K3Moonshot AI | 1223 | 587 | 62.3 | $15.00 | |
| 17 | Kimi K2.5Moonshot AI | 1219 | Not published | 38.6 | $2.85 | |
| 18 | Qwen3.7 MaxQwen | 1187 | 1,213 | 54.1 | $4.425 | |
| 19 | GLM 4.6Z.ai | 1180 | Not published | 8.8 | $2.20 | |
| 20 | GPT-5.5OpenAI | 1173 | 1,226 | 71.3 | $30.00 | |
| 21 | Gemini 3 Flash PreviewGoogle | 1161 | Not published | 43.7 | $3.00 | |
| 22 | Kimi K2.7 CodeMoonshot AI | 1147 | 1,103 | 53.4 | $4.00 | |
| 23 | GLM 5Z.ai | 1144 | Not published | 41.7 | $3.20 | |
| 24 | GPT-5.2OpenAI | 1142 | Not published | 55.9 | $14.00 | |
| 25 | GPT-5.2-CodexOpenAI | 1142 | Not published | 45.4 | $14.00 | |
| 26 | GLM 5.2Z.ai | 1142 | Not published | 61.1 | $4.40 | |
| 27 | GPT-5.4OpenAI | 1135 | Not published | 64.4 | $15.00 | |
| 28 | GPT-5.3-CodexOpenAI | 1124 | Not published | 60.0 | $14.00 | |
| 29 | Kimi K2.6Moonshot AI | 1120 | 1,497 | 52.1 | $4.00 | |
| 30 | Muse Spark 1.1Meta | 1103 | Not published | 70.3 | $4.25 | |
| 31 | Gemini 3.5 FlashGoogle | 1098 | 1,345 | 39.8 | $9.00 | |
| 32 | Grok 4.20xAI | 1089 | Not published | 33.3 | $2.50 | |
| 33 | DeepSeek V4 Pro 0423DeepSeek | 1059 | Not published | 38.7 | $2.60 | |
| 34 | GLM 4.7Z.ai | 1058 | Not published | 55.1 | $2.20 | |
| 35 | GLM 5.1Z.ai | 1053 | 431 | 49.2 | $4.40 | |
| 36 | GLM 5V TurboZ.ai | 957 | 1,944 | Not benchmarked | $4.00 | |
| 37 | Grok 4.3xAI | 914 | 314 | 37.4 | $2.50 |
Only models in our catalog are listed, and each appears once with its best published run. 146 rows were read from the source on October 22, 2026, and rows that match no catalog model are left out. Every other column is a figure the board publishes for the same run.