Vibe Code Bench 1-100 (Vals AI) leaderboard
- Claude Opus 5.5 leads
- 30.4%
- models measured
- 23
- lead over #2
- +1.8 pts
- median
- 18.2%
Leader on Vibe Code Bench 1-100 (Vals AI) · read October 22
# Vibe Code Bench 1-100 (Vals AI), read October 22
leaders = {
"Claude Opus 5.5": 30.4%, "Claude Opus 5": 28.5%, "Claude Fable 5.1": 28.0%,
}
On October 22, 2026, 23 of our catalog's models had a Vibe Code Bench 1-100 (Vals AI) result, and Anthropic Claude Opus 5.5 led with 30.4%, ahead of Anthropic Claude Opus 5 at 28.5%.
An agent extends a working web app over a series of requests, and each step is checked for regressions. The score as Vals AI publishes it.
Role in the index
A coding board we show beside the Vibe Coding Index, not inside it: the index keeps the six boards it was defined on, so a score does not move every time a board is added. Read the methodology
| # | Model | Result | Cost per test | Duration, as printed | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5Anthropic | 30.4% | $44.12 | 2h 28m | 88.8 | $20.00 | |
| 2 | Claude Opus 5Anthropic | 28.5% | $41.49 | 1h 22m | 83.0 | $25.00 | |
| 3 | Claude Fable 5.1Anthropic | 28.0% | $149.66 | 2h 53m | 84.2 | $50.00 | |
| 4 | GPT-6 AstraOpenAI | 27.6% | $84.72 | 3h 26m | 85.1 | $50.00 | |
| 5 | Muse WonderMeta | 26.6% | $13.72 | 31m 08s | 71.9 | $20.00 | |
| 6 | GPT-5.6 LunaOpenAI | 22.6% | $4.91 | 2h 03m | 58.0 | $1.20 | |
| 7 | Gemini 4 ArgonGoogle | 21.2% | $18.66 | 1h 43m | 81.0 | $10.00 | |
| 8 | Claude Haiku 5.5Anthropic | 20.4% | $8.30 | 4h 06m | 70.7 | $0.50 | |
| 9 | GPT-5.6 SolOpenAI | 20.0% | $47.43 | 2h 37m | 73.5 | $20.00 | |
| 10 | GLM 5.3Z.ai | 20.0% | $8.24 | 1h 26m | 70.6 | $4.40 | |
| 11 | Gemini 3.8 FlashGoogle | 18.8% | $18.91 | 1h 06m | 61.6 | $3.75 | |
| 12 | Kimi K3Moonshot AI | 18.2% | $9.40 | 1h 48m | 62.3 | $15.00 | |
| 13 | DeepSeek V4 Pro 0813DeepSeek | 17.5% | $5.89 | 37m 42s | 56.1 | $1.98 | |
| 14 | DeepSeek V4.1 FlashDeepSeek | 16.4% | $0.69 | 28m 21s | 62.5 | $0.84 | |
| 15 | GLM 5.3 FlashZ.ai | 16.0% | $0.46 | 39m 03s | 57.7 | $0.50 | |
| 16 | GPT-5.6 TerraOpenAI | 14.8% | $33.33 | 1h 49m | 63.9 | $12.00 | |
| 17 | Grok 4.6xAI | 14.8% | $26.27 | 2h 11m | 65.7 | $6.00 | |
| 18 | Mistral Large 4Mistral AI | 14.3% | $6.75 | 1h 00m | 56.4 | $4.18 | |
| 19 | Claude Sonnet 5Anthropic | 13.8% | $71.15 | 4h 16m | 57.7 | $10.00 | |
| 20 | Qwen3.8 Max (0902)Qwen | 12.8% | $9.68 | 1h 36m | 62.4 | $6.00 | |
| 21 | MiniMax M3MiniMax | 9.2% | $3.03 | 26m 10s | 39.2 | $1.20 | |
| 22 | T InklingThinking Machines | 7.3% | $5.00 | 50m 02s | 31.5 | $4.05 | |
| 23 | Gemini 3.1 Pro PreviewGoogle | 6.7% | $3.36 | 11m 28s | 32.3 | $12.00 |
Only models in our catalog are listed, and each appears once with its best published run. 137 rows were read from the source on October 22, 2026, and rows that match no catalog model are left out. Cost per test is what Vals AI reports one test cost on the run behind the result, in US dollars. Every other column is a figure the board publishes for the same run.