Inference
EN
Vals AI

HLAB-Vals-1.1

Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools

ModelsModels with a result.35
Top resultBest result on this benchmark.27.1%Gemini 4 Argon
Top-3 spreadPoints from first to third.2.1 pts
UpdatedDate the source changed the results.9 Nov 2026

Result and cost

34 of 35 results

Results

35 results
#ModelResultThe score from the source.CostUS dollars to run one task.TimeTime to run one task.
01Gemini 4 Argon27.1% ±3.7$10.5529 min
02Muse Spark Open25.0% ±3.6$13.634.0 h
02Muse Ember25.0% ±3.7$0.1186 min
04Muse Wonder23.8% ±3.6$7.0815 min
05Muse Wonder Full23.3% ±3.5$5.4144 min
06Inkling 2 Small21.7% ±3.3$1.2322 min
07Grok 4.718.3% ±3.2$10.7536 min
08Gemini 4 Carbon17.5% ±3.0$6.5229 min
08Mistral Large 417.5% ±3.2$2.5558 min
10Mistral Large 4 OSS15.8% ±3.1$0.73311 min
11Qwen3.8 Max Internal15.4% ±3.1$3.3649 min
12GLM-5.314.6% ±2.9$3.9932 min
13MiMo-V2.6-Pro13.8% ±2.8$0.21731 min
14MiMo-V2.6-Flash13.3% ±2.8$0.08622 min
15Step 5 Preview12.9% ±2.8$1.1317 min
15Kimi K312.9% ±2.8$4.1443 min
17Hy412.5% ±2.7$1.0419 min
18Gemini 3.8 Flash12.1% ±2.6$3.6619 min
19DeepSeek V4 Pro 081310.8% ±2.7$0.80211 min
20Claude Sonnet 5.510.4% ±2.6$17.7173 min
20GLM-5.3-Flash10.4% ±2.6$0.32312 min
22Claude Opus 5.59.2% ±2.4$21.8158 min
23Hy4 preview8.8% ±2.2$0.98433 min
24MiniMax M3.1 Flash7.5% ±2.3$0.47325 min
25GPT-6 Astra7.1% ±1.9$25.4633 min
26GPT-6.1 Sol6.3% ±1.8$3.5440 min
27Claude Fable 5.25.0% ±1.7$19.3733 min
27Mercury 35.0% ±1.6$0.18815 min
29Beam 1.04.6% ±1.8—2 min
30Claude Haiku 5.54.2% ±1.5$4.2644 min
30GPT-6 Luna4.2% ±1.6$0.29822 min
32Inkling3.8% ±1.5$1.259 min
32Nemotron 3.5 Super VL 120B A12B3.8% ±1.3$0.3733 min
34Inkling-Small2.9% ±1.4$0.1798 min
35GPT-6.1 Terra1.7% ±1.0$1.4814 min