Vals Finance Index v1 (Vals Finance Index)
We show this table for reference; we do not rank on it.
Vals AI composite of agentic finance work: analyst research on company filings, financial modeling in Excel and corporate tax research, weighted by each sub-sector's share of U.S. GDP.
Evaluation results published by Vals AI. Published leaderboard aggregates and evaluation settings only. Dataset contents are not included.
GDP-weighted accuracy on Vals Finance Index — October 27, 2026
We mirror the published gdp-weighted accuracy view for Vals Finance Index. Gemini 4 Carbon leads the public snapshot at 66.35%, followed by Claude Opus 5.5 (65.59%) and Gemini 4 Argon (64.89%). We do not use these results to rank models overall.
Gemini 4 Carbon
high reasoning
Claude Opus 5.5
Anthropic
Gemini 4 Argon
high reasoning
33 modelsAgenticCurrentDisplay onlyUpdated October 27, 2026
GDP-weighted accuracy table (33 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Gemini 4 CarbonGooglehigh reasoning | 66.35% | Not reported | Closed |
| 2 | Claude Opus 5.5Anthropic | 65.59% | Not reported | Closed |
| 3 | Gemini 4 ArgonGooglehigh reasoning | 64.89% | Not reported | Closed |
| 4 | Muse Wonder FullMetamax reasoning | 64.55% | Not reported | Closed |
| 5 | Claude Sonnet 5.5Anthropic | 64.53% | Not reported | Closed |
| 6 | Qwen3.8 Max InternalAlibaba | 61.66% | Not reported | Unknown |
| 7 | Muse WonderMetamax reasoning | 61.23% | Not reported | Closed |
| 8 | DeepSeek V4.1 Flash 1017DeepSeekmax reasoning | 60.89% | Not reported | Open |
| 9 | GLM-5.3Z.AImax reasoning | 60.54% | Not reported | Open |
| 10 | Gemini 3.8 FlashGooglehigh reasoning | 60.46% | Not reported | Closed |
| 11 | Grok 4.7xAIxhigh reasoning | 58.46% | Not reported | Closed |
| 12 | Kimi K3Moonshot AImax reasoning | 57.95% | Not reported | Open |
| 13 | MiMo-V2.6-ProXiaomi | 57.85% | Not reported | Open |
| 14 | Hy4 previewTencent | 57.51% | Not reported | Open |
| 15 | Claude Haiku 5.5Anthropic | 57.45% | Not reported | Closed |
| 16 | MiMo-V2.6-FlashXiaomi | 57.15% | Not reported | Open |
| 17 | DeepSeek V4.1 Flash 1017DeepSeekmax reasoning | 57.08% | Not reported | Open |
| 18 | Mistral Large 4Mistralhigh reasoning | 56.90% | Not reported | Open |
| 19 | Kimi K3.1 ProMoonshot AImax reasoning | 56.01% | Not reported | Pending |
| 20 | MiniMax M3.1 FlashMiniMax | 55.91% | Not reported | Closed |
| 21 | GLM-5.3-FlashZ.AImax reasoning | 55.83% | Not reported | Open |
| 22 | GPT-6.1 SolOpenAImax reasoning | 55.81% | Not reported | Closed |
| 23 | GPT-6 AstraOpenAImax reasoning | 55.64% | Not reported | Closed |
| 24 | Step 5 PreviewStepFun | 55.43% | Not reported | Open |
| 25 | GPT-6 LunaOpenAImax reasoning | 53.05% | Not reported | Closed |
| 26 | Muse Spark OpenMetaxhigh reasoning | 52.99% | Not reported | Open |
| 27 | DeepSeek V4 Pro 0813DeepSeekmax reasoning | 52.94% | Not reported | Open |
| 28 | GPT-6.1 TerraOpenAImax reasoning | 52.86% | Not reported | Closed |
| 29 | MiniMax M3MiniMax | 48.37% | Not reported | Open |
| 30 | Mercury 3Inceptionhigh reasoning | 47.22% | Not reported | Closed |
| 31 | Beam 1.0Reflection | 46.50% | Not reported | Unknown |
| 32 | InklingThinking Machines Lab0.99 reasoning | 38.25% | Not reported | Open |
| 33 | Inkling-SmallThinking Machines Lab0.99 reasoning | 36.27% | Not reported | Open |
How to read this leaderboard
Operator receipt: 33 sourced rows are currently displayable on this page; the leading published row is Gemini 4 Carbon at 66.35%.
Honest limit: It averages three Vals benchmarks with private task sets, so a model's index moves with any one component, and a model is scored only when it has all three. The GDP weights make tax count for about a quarter of the score. It measures model-and-agent runs on finance tasks, not professional practice.
How Vals Finance Index is shown here
We mirror Vals AI's Finance Index v1, released October 21, 2026. It scores agentic finance work on its own: Finance Agent v2.1 (analyst research on public company filings) and the Excel Modeling Benchmark v1.1 (building and editing financial models) form the securities and investments sub-sector, and Tax Agent Bench (corporate tax research with cited authority) forms the tax sub-sector.
Vals weights the sub-sectors by their share of U.S. GDP, 1.6 for securities and investments and 0.5 for tax, using Bureau of Economic Analysis value added. A model is scored only when all three components have a result. The results stay display only and do not enter model rankings.
Snapshot
The published Vals Finance Index snapshot places Gemini 4 Carbon first at 66.35%. The third row is 1.46 points behind. The broader top-10 range is 5.89 points, so many of the published results sit in a relatively narrow band.
33 models have been evaluated on Vals Finance Index. The benchmark falls in the Agentic category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. Vals Finance Index is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About Vals Finance Index
Year
2026
Tasks
3 finance benchmarks
Format
GDP-weighted accuracy across three benchmarks
Difficulty
Agentic analyst research, Excel financial modeling and corporate tax research
Vals released v1 on October 21, 2026. Finance Agent v2.1 and the Excel Modeling Benchmark v1.1 average into a securities and investments score (weight 1.6) and Tax Agent Bench is the tax score (weight 0.5), combined as (1.6 x securities + 0.5 x tax) / 2.1. Weights come from Bureau of Economic Analysis value added as a share of U.S. GDP. We mirror the published aggregates as a display-only composite, outside overall and category rankings.
Freshness and provenance
Version
Vals Finance Index 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does Vals Finance Index measure?
Vals AI composite of agentic finance work: analyst research on company filings, financial modeling in Excel and corporate tax research, weighted by each sub-sector's share of U.S. GDP.
Which model leads the published Vals Finance Index snapshot?
Gemini 4 Carbon currently leads the published Vals Finance Index snapshot with 66.35% gdp-weighted accuracy. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on Vals Finance Index?
The October 27, 2026 snapshot contains 33 AI models.
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 6,100+ subscribers.
One email each week. Unsubscribe anytime.