Proprietary

Vals Index

Updated 8/11/2026

A single measure of AI's potential economic impact — model performance across finance and coding tasks, weighted by each sector's contribution to U.S. GDP.

As of August 11, 2026, Claude Fable 5 ranks first on the Vals Index v1.2 with 75.14%, followed by Claude Opus 5 (74.82%) and Kimi K3 (74.70%).

Vals Index
ACCURACY

Vals Index leaderboard

Rank Model Accuracy Cost / Test Input / Output Cost Duration
1 Claude Fable 5 75.14% $11.00 $10 / $50 17m31s
2 Claude Opus 5 74.82% $8.54 $5 / $25 19m42s
3 Kimi K3 74.70% $2.34 $3 / $15 20m23s
4 GPT-5.6 Sol 73.12% $7.46 $4 / $20 12m33s
5 Muse Spark 1.2 71.88% $0.70 $1.25 / $4.25 10m10s
6 Claude Opus 4.8 70.36% $7.52 $5 / $25 22m11s
7 GPT-5.6 Luna 69.88% $0.22 $0.2 / $1.2 9m14s
8 Claude Sonnet 5 68.61% $9.01 $2 / $10 22m04s
9 Muse Spark 1.1 68.41% $0.50 $1.25 / $4.25 6m29s
10 GPT 5.5 67.95% $4.60 $5 / $30 11m24s
11 Claude Opus 4.7 66.10% $6.14 $5 / $25 12m42s
12 Qwen 3.8 Max 65.47% $2.68 $2 / $6 45m59s
13 Grok 4.5 65.30% $1.25 $2 / $6 6m28s
14 GPT-5.6 Terra 65.14% $1.11 $2 / $12 5m52s
15 GLM 5.2 65.02% $2.08 $1.4 / $4.4 18m42s
16 Gemini 3.6 Flash 64.92% $1.42 $1.5 / $7.5 12m24s
17 DeepSeek V4 Flash 0731 63.95% $0.06 $0.44 / $1.32 14m20s
18 Gemini 3.5 Flash 62.75% $1.40 $1.5 / $9 6m04s
19 Claude Sonnet 4.6 60.06% $2.15 $3 / $15 11m52s
20 MiniMax-M3 58.94% $1.50 $0.6 / $2.4 24m38s
21 Qwen 3.7 Max 57.49% $5.22 $2.5 / $7.5 30m32s
22 DeepSeek V4 55.62% $0.83 $1.32 / $3.96 22m07s
23 Kimi K2.6 55.17% $0.71 $0.95 / $4 19m49s
24 Gemini 3.1 Pro Preview (02/26) 53.77% $1.35 $2 / $12 7m15s
25 GLM 5.1 52.45% $0.86 $1 / $3.2 14m59s
26 GPT 5.4 Mini 52.42% $0.66 $0.75 / $4.5 17m30s
27 Qwen 3.7 Plus 52.33% $0.35 $0.4 / $1.6 13m53s
28 MiMo V2.5 51.57% $0.04 $0.14 / $0.28 9m45s
29 Gemini 3.5 Flash Lite 51.00% $0.42 $0.3 / $2.5 5m42s
30 MiMo V2.5 Pro 50.74% $0.09 $0.435 / $0.87 13m59s
31 Inkling Small 50.37% $0.19 $0.3 / $1.2 15m29s
32 Gemini 3 Flash (12/25) 49.55% $0.40 $0.5 / $3 5m53s
33 Inkling 49.28% $0.84 $1 / $4.05 20m47s
34 Qwen 3.6 Plus 48.89% $1.55 $0.5 / $3 13m59s
35 GPT 5.4 Nano 46.63% $0.32 $0.2 / $1.25 15m30s
36 Grok 4.3 46.48% $0.69 $1.25 / $2.5 6m07s
37 Nemotron 3 Ultra 43.99% N/A N/A 10m38s
38 MiniMax-M2.7 41.56% $0.72 $0.3 / $1.2 13m03s
39 Claude Haiku 4.5 (Thinking) 40.90% $0.56 $1 / $5 6m17s
40 Grok 4.20 (Reasoning) 39.50% $0.54 $2 / $6 3m44s
41 Ling 3.0 Flash 38.02% $0.10 $0.075 / $0.22 6m42s
42 Mistral Medium 3.5 37.81% $4.82 $1.5 / $7.5 16m44s
43 Gemini 3.1 Flash Lite Preview 36.20% $0.17 $0.25 / $1.5 3m29s
44 Laguna M.1 33.33% N/A N/A 20m12s
45 Laguna XS.2 30.04% N/A N/A 12m27s

Results

Industry Average Accuracy Comparison

Methodology

This archived version scores finance and coding only, on index-specific subsets of its components, before v2 re-added a legal sector and re-sourced every component from its benchmark’s published standalone run.

Finance (weight: 2.0): ~$2T contribution to U.S. GDP

Coding (weight: 1.4): ~$1.4T contribution to U.S. GDP

Coding = 0.25 * SWE_Bench + 0.25 * TBench + 0.5 * VibeCodeBench
Vals_Index = (2.0 * AVG(CorpFin, FinanceAgent) + 1.4 * Coding) / 3.4