Excel Modeling Benchmark (Vals AI) leaderboard
- Claude Fable 5.1 leads
- 76.7%
- models measured
- 73
- lead over #2
- +0.7 pts
- median
- 57.9%
Leader on Excel Modeling Benchmark (Vals AI) · read October 22
On October 22, 2026, 73 of our catalog's models had a Excel Modeling Benchmark (Vals AI) result, and Anthropic Claude Fable 5.1 led with 76.7%, ahead of Anthropic Claude Opus 5.5 at 75.9%.
Financial models built in spreadsheets by an agent, graded against set criteria. Share correct, as Vals AI grades it.
Role in the index
Excel Modeling Benchmark (Vals AI) is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Cost per test | Duration, as printed | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|---|---|
| 1 | Claude Fable 5.1Anthropic | 76.7% | $15.93 | 29m 32s | 84.2 | $50.00 | |
| 2 | Claude Opus 5.5Anthropic | 75.9% | $11.02 | 35m 06s | 88.8 | $20.00 | |
| 3 | Claude Sonnet 5.5Anthropic | 75.7% | $8.79 | 36m 45s | 87.3 | $10.00 | |
| 4 | Gemini 4 ArgonGoogle | 75.2% | $11.80 | 38m 26s | 81.0 | $10.00 | |
| 5 | Claude Fable 5Anthropic | 73.7% | $12.35 | 26m 34s | 79.3 | $50.00 | |
| 6 | Claude Opus 5Anthropic | 73.6% | $6.07 | 25m 59s | 83.0 | $25.00 | |
| 7 | GPT-5.6 SolOpenAI | 72.3% | $6.01 | 9m 50s | 73.5 | $20.00 | |
| 8 | Gemini 3.8 FlashGoogle | 72.2% | $8.24 | 12m 49s | 61.6 | $3.75 | |
| 9 | GPT-6 AstraOpenAI | 71.7% | $5.83 | 13m 24s | 85.1 | $50.00 | |
| 10 | GPT-6 SolOpenAI | 71.5% | $1.28 | 10m 22s | 76.0 | $10.00 | |
| 11 | Gemini 3.7 FlashGoogle | 71.3% | $5.69 | 10m 51s | 58.0 | $3.75 | |
| 12 | GPT-6.1 SolOpenAI | 70.8% | $1.14 | 18m 28s | 82.4 | $10.00 | |
| 13 | Claude Opus 4.8Anthropic | 69.4% | $12.06 | 43m 52s | 65.0 | $25.00 | |
| 14 | GPT-6 LunaOpenAI | 68.5% | $0.14 | 17m 21s | 57.3 | $0.50 | |
| 15 | GPT-5.6 LunaOpenAI | 67.1% | $0.37 | 11m 00s | 58.0 | $1.20 | |
| 16 | Muse WonderMeta | 67.1% | $5.60 | 17m 14s | 71.9 | $20.00 | |
| 17 | Grok 4.7xAI | 67.0% | $6.48 | 34m 57s | 66.7 | $6.00 | |
| 18 | Kimi K3Moonshot AI | 66.7% | $3.07 | 12m 21s | 62.3 | $15.00 | |
| 19 | Claude Sonnet 5Anthropic | 66.3% | $10.29 | 49m 46s | 57.7 | $10.00 | |
| 20 | GPT-5.6 TerraOpenAI | 66.2% | $2.19 | 13m 33s | 63.9 | $12.00 | |
| 21 | MiMo-V2.6-FlashXiaomi | 65.5% | $0.11 | 25m 22s | 54.7 | $0.28 | |
| 22 | Gemini 3.6 FlashGoogle | 65.4% | $3.89 | 12m 06s | 62.5 | $3.75 | |
| 23 | GPT-5.5OpenAI | 64.5% | $3.27 | 15m 11s | 71.3 | $30.00 | |
| 24 | Claude Opus 4.7Anthropic | 63.7% | $12.76 | 38m 36s | 67.5 | $25.00 | |
| 25 | Gemini 3.5 FlashGoogle | 63.5% | $5.17 | 18m 44s | 39.8 | $9.00 | |
| 26 | Ember-1Fireworks | 63.4% | $2.46 | 31m 07s | 52.6 | $15.00 | |
| 27 | MiMo-V2.6-ProXiaomi | 62.9% | $0.33 | 55m 14s | 65.4 | $0.87 | |
| 28 | Grok 4.6xAI | 62.7% | $3.06 | 52m 59s | 65.7 | $6.00 | |
| 29 | Muse Spark 1.3Meta | 62.7% | $3.55 | 22m 55s | 57.4 | $4.25 | |
| 30 | DeepSeek V4.1 Flash 1017DeepSeek | 61.8% | $0.36 | 16m 47s | 70.3 | $0.60 | |
| 31 | GLM 5.2Z.ai | 61.5% | $4.47 | 58m 52s | 61.1 | $4.40 | |
| 32 | Claude Sonnet 4.6Anthropic | 60.2% | $7.23 | 54m 16s | 57.3 | $15.00 | |
| 33 | Qwen3.8 Max (0902)Qwen | 60.1% | $2.66 | 57m 47s | 62.4 | $6.00 | |
| 34 | Qwen3.8 27BQwen | 59.7% | $4.56 | 22m 51s | 74.2 | $2.55 | |
| 35 | Claude Haiku 5.5Anthropic | 59.0% | $1.15 | 18m 03s | 70.7 | $0.50 | |
| 36 | Hy4 previewTencent | 58.1% | $0.79 | 30m 39s | 44.3 | $2.501 | |
| 37 | Kimi K2.6Moonshot AI | 57.9% | $2.20 | 39m 48s | 52.1 | $4.00 | |
| 38 | DeepSeek V4.1 FlashDeepSeek | 57.2% | $0.23 | 13m 00s | 62.5 | $0.84 | |
| 39 | Step 5 PreviewStepFun | 57.1% | $1.33 | 43m 35s | 61.1 | $2.70 | |
| 40 | Qwen3.7 MaxQwen | 57.0% | $2.86 | 32m 37s | 54.1 | $4.425 | |
| 41 | DeepSeek V4 Flash 0731DeepSeek | 57.0% | $0.38 | 26m 14s | 51.8 | $0.396 | |
| 42 | Muse Spark 1.2Meta | 57.0% | $2.31 | 17m 07s | 55.7 | $4.25 | |
| 43 | Muse Spark 1.1Meta | 56.4% | $1.11 | 14m 52s | 70.3 | $4.25 | |
| 44 | GLM 5.3Z.ai | 56.3% | $3.79 | 36m 12s | 70.6 | $4.40 | |
| 45 | Mistral Large 4Mistral AI | 56.2% | $4.84 | 1h 22m | 56.4 | $4.18 | |
| 46 | GLM 5.3 FlashZ.ai | 55.9% | $0.71 | 1h 15m | 57.7 | $0.50 | |
| 47 | MiMo-V2.5-ProXiaomi | 55.2% | $0.22 | 29m 08s | 31.6 | $0.9152 | |
| 48 | MiMo-V2.5Xiaomi | 55.1% | $0.08 | 30m 38s | 50.0 | $0.336 | |
| 49 | Muse Spark OpenMeta | 53.7% | $5.35 | 1h 25m | 47.4 | $4.25 | |
| 50 | Grok 4.5xAI | 52.9% | $1.55 | 18m 28s | 58.6 | $6.00 | |
| 51 | DeepSeek V4 Pro 0813DeepSeek | 52.8% | $1.08 | 37m 27s | 56.1 | $1.98 | |
| 52 | Gemini 3.1 Pro PreviewGoogle | 52.6% | $3.77 | 11m 49s | 32.3 | $12.00 | |
| 53 | DeepSeek V4 Pro 0423DeepSeek | 51.6% | $0.80 | 13m 42s | 38.7 | $2.976 | |
| 54 | Qwen3.7 PlusQwen | 49.3% | $0.60 | 36m 49s | 33.2 | $1.28 | |
| 55 | MiniMax M3MiniMax | 47.8% | $2.09 | 31m 30s | 39.2 | $1.20 | |
| 56 | GPT-5.4 MiniOpenAI | 45.4% | $2.15 | 43m 21s | 35.8 | $4.50 | |
| 57 | GPT-5.4 NanoOpenAI | 44.8% | $0.57 | 29m 27s | 44.8 | $1.25 | |
| 58 | Gemini 3.5 Flash LiteGoogle | 43.2% | $1.23 | 8m 14s | 49.0 | $2.50 | |
| 59 | T InklingThinking Machines | 38.4% | $1.67 | 26m 22s | 31.5 | $4.05 | |
| 60 | Qwen3.6 PlusQwen | 32.9% | $2.87 | 31m 14s | 44.1 | $1.95 | |
| 61 | T Inkling SmallThinking Machines | 31.8% | $0.25 | 19m 34s | 27.3 | $1.20 | |
| 62 | Nemotron 3 UltraNVIDIA | 31.7% | Not published | 39m 42s | 30.0 | $2.40 | |
| 63 | Kimi K2.5Moonshot AI | 28.5% | $0.79 | 1h 03m | 38.6 | $2.85 | |
| 64 | MiniMax M2.7MiniMax | 28.4% | $1.66 | 20m 07s | 36.2 | $1.20 | |
| 65 | Gemini 3 Flash PreviewGoogle | 26.0% | $0.45 | 5m 29s | 43.7 | $3.00 | |
| 66 | Ling 3.0 FlashinclusionAI | 25.0% | $0.18 | 12m 15s | 29.6 | $0.1215 | |
| 67 | Claude Haiku 4.5Anthropic | 22.7% | $1.03 | 11m 59s | 35.6 | $5.00 | |
| 68 | Grok 4.3xAI | 18.4% | $0.63 | 4m 32s | 37.4 | $2.50 | |
| 69 | Nemotron 3.5 LightningNVIDIA | 17.0% | $0.18 | 39m 19s | 22.6 | $0.147 | |
| 70 | Grok 4.20xAI | 11.7% | $0.47 | 6m 08s | 33.3 | $2.50 | |
| 71 | Mistral Medium 3.5Mistral AI | 11.2% | $10.60 | 51m 22s | 25.5 | $7.50 | |
| 72 | Mercury 2.5Inception | 8.8% | $0.42 | 5m 01s | 1.6 | $0.15 | |
| 73 | Gemini 3.1 Flash Lite PreviewGoogle | 8.6% | $0.07 | 1m 07s | 22.7 | $1.50 |
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.