General intelligence

Vals Index

GDP-weighted agent performance across finance, coding and legal work

What it measures

Vals Index is Vals AI’s composite measure of agent performance on real-world work, currently at version 2. It covers finance, coding and legal tasks: financial research and Excel modeling; terminal work, app building and code migration; and legal research and long-horizon legal work products.

Think of it as a report card with differently weighted subjects: average each sector’s benchmark scores first, then combine the three sector averages. Finance has the greatest weight, so models with similar coding results can have different overall scores because of financial research and modeling.

Finance comprises Finance Agent v2 and the Excel Modeling Benchmark; coding comprises Terminal-Bench 2.1, Vibe Code Bench and Code Migration; legal comprises Legal Research Bench and Harvey’s Legal Agent Benchmark. The weights use the sectors’ approximate shares of U.S.

GDP: finance and insurance 8.0, information 5.6, and legal services 1.2, totaling 14.8; normalized weights are approximately 54.1%, 37.8% and 8.1%.

Reading the scores

Higher scores mean stronger overall performance under this specific task mix and weighting. The formula is (8 × Finance + 5.6 × Coding + 1.2 × Legal) ÷ 14.8. For example, sector scores of 60, 80 and 40 yield approximately 65.9%. This is neither a simple success rate across all tasks nor a percentage of jobs automated or GDP generated. This snapshot includes all 57 model results from the v2 leaderboard marked updated September 11, 2026, retaining published reasoning settings. Components and weights change between versions, so their scores should not be compared directly. Code Migration uses a fixed subset of 50 CLI migration tasks plus all 10 COBOL tasks, weighted 75% / 25%; its full standalone leaderboard score cannot simply be substituted. We retain the published composite scores instead of recomputing them from other benchmarks on this site, and do not present source standard errors as 95% confidence intervals.

SourceVals AI

Explore the finance component: Finance Agent v2

Model performance

Checked September 16, 202657 results

Vals Index — Model performance. Select a column heading to sort.

Claude Fable 5.1 (max)Anthropic	68.8%¹
Claude Opus 5 (max)Anthropic	67.2%²
GPT-6 Astra (max)OpenAI	66.6%³
Claude Fable 5 (max)Anthropic	66.0%⁴
Muse Spark 1.3 Max (max)Meta	64.5%
GPT-5.6 Sol (max)OpenAI	63.7%
Gemini 3.8 Flash (high)Google	62.3%
Claude Opus 4.8 (max)Anthropic	60.9%
Muse Spark 1.3 (xhigh)Meta	60.3%
GPT-5.6 Luna (max)OpenAI	59.9%⁵
Claude Sonnet 5 (max)Anthropic	59.6%
GPT-5.6 Terra (max)OpenAI	59.6%⁶
Gemini 3.7 Flash (high)Google	59.3%
Grok 4.6 (high)xAI	59.2%
DeepSeek V4.1 Flash (high)DeepSeek	57.9%
Kimi K3Moonshot AI	57.8%
GPT-5.5 (xhigh)OpenAI	57.4%
Muse Spark 1.2 (xhigh)Meta	57.1%
GLM-5.3 (max)Z AI	57.0%
Claude Opus 4.7 (high)Anthropic	56.1%
Gemini 3.6 Flash (high)Google	55.4%
Muse Spark 1.1 (xhigh)Meta	54.8%
DeepSeek V4 Flash 0731 (high)DeepSeek	53.6%
GLM-5.2Z AI	53.1%
Gemini 3.5 Flash (high)Google	53.1%
DeepSeek V4 Pro 0813 (max)DeepSeek	52.4%
Qwen 3.8 MaxAlibaba	51.8%
Grok 4.5 (high)xAI	51.5%
Claude Sonnet 4.6 (max)Anthropic	50.6%
Qwen 3.8 27B (xhigh)Alibaba	48.5%
GLM-5.3 Flash (max)Z AI	47.2%
Qwen 3.7 MaxAlibaba	44.8%
Kimi K2.6Moonshot AI	43.5%
DeepSeek V4 Pro (max)DeepSeek	42.9%
MiniMax-M3MiniMax	42.7%
Gemini 3.1 Pro Preview (02/26) (high)Google	41.9%
MiMo V2.5 ProXiaomi	41.0%
MiMo V2.5Xiaomi	39.9%
GPT 5.4 Mini (xhigh)OpenAI	39.6%
Qwen 3.7 PlusAlibaba	38.6%
Gemini 3.5 Flash Lite (high)Google	36.7%
Inkling (0.99)Thinking Machines	34.1%
GPT 5.4 Nano (high)OpenAI	33.0%
Qwen 3.6 PlusAlibaba	32.0%
Inkling Small (0.99)Thinking Machines	31.9%
Gemini 3 Flash (12/25) (high)Google	29.4%
Nemotron 3 UltraNVIDIA	27.4%
Kimi K2.5Moonshot AI	26.3%
MiniMax-M2.7MiniMax	24.6%
Grok 4.3 (high)xAI	24.3%
Claude Haiku 4.5 (Thinking)Anthropic	22.9%
Ling 3.0 FlashAnt Group	21.7%
Mistral Medium 3.5 (high)Mistral	17.9%
Grok 4.20 (Reasoning)xAI	17.6%
Gemini 3.1 Flash Lite Preview (high)Google	15.5%
Mercury 2.5 (high)Inception	13.0%
Nemotron 3.5 LightningNVIDIA	11.5%