Design Arena: spreadsheets leaderboard
- models measured
- 20
- lead over #2
- +4 Elo
- median
- 1185
Leader on Design Arena: spreadsheets · read October 21
On October 21, 2026, 20 of our catalog's models had a Design Arena: spreadsheets result, and Anthropic Claude Fable 5.1 led with 1365, ahead of OpenAI GPT-6 Astra at 1361.
Design Arena shows people two designs made by two models from the same prompt (a spreadsheet, 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: spreadsheets is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Battles | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|---|
| 1 | Claude Fable 5.1Anthropic | 1365 | 159 | 84.2 | $50.00 | |
| 2 | GPT-6 AstraOpenAI | 1361 | 165 | 85.1 | $50.00 | |
| 3 | Claude Opus 5.5Anthropic | 1352 | 172 | 88.8 | $20.00 | |
| 4 | GLM 5.1Z.ai | 1315 | Not published | 49.2 | $4.40 | |
| 5 | Kimi K2.6Moonshot AI | 1276 | Not published | 52.1 | $4.00 | |
| 6 | Claude Opus 4.8Anthropic | 1270 | 20 | 65.0 | $25.00 | |
| 7 | Claude Sonnet 5.5Anthropic | 1244 | 116 | 87.3 | $10.00 | |
| 8 | GPT-6.1 SolOpenAI | 1242 | 155 | 82.4 | $10.00 | |
| 9 | DeepSeek V4.1 Flash 1017DeepSeek | 1230 | 164 | 70.3 | $0.60 | |
| 10 | Muse WonderMeta | 1227 | 153 | 71.9 | $20.00 | |
| 11 | Gemini 4 ArgonGoogle | 1142 | 58 | 81.0 | $10.00 | |
| 12 | Kimi K3Moonshot AI | 1058 | 142 | 62.3 | $15.00 | |
| 13 | Gemini 3.8 FlashGoogle | 1056 | 102 | 61.6 | $3.75 | |
| 14 | Gemini 3.7 FlashGoogle | 1053 | 43 | 58.0 | $3.75 | |
| 15 | Muse Spark 1.3Meta | 1043 | 163 | 57.4 | $4.25 | |
| 16 | Gemini 3.6 FlashGoogle | 1034 | 51 | 62.5 | $3.75 | |
| 17 | Claude Sonnet 5Anthropic | 1021 | 58 | 57.7 | $10.00 | |
| 18 | MiniMax M3MiniMax | 1009 | 47 | 39.2 | $1.20 | |
| 19 | Muse Spark 1.1Meta | 975 | Not published | 70.3 | $4.25 | |
| 20 | Qwen3.8 Max (0902)Qwen | 937 | 38 | 62.4 | $6.00 |
Only models in our catalog are listed, and each appears once with its best published run. 143 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.