Benchmarks · composite
Vals Index
GDP-weighted measure of agentic model performance across finance, coding, and legal tasks.
21 models · 21 results · updated 2026-10-27 · vals.ai ↗
Leaderboard
Sonnet 5.5 max
More from this benchmark
1 published measuresTime per task29 results · lower is better
All results
21 rows| # | Model | Effort | Vals Index | Subscription $ / task | API $ / task | Time |
|---|---|---|---|---|---|---|
| 1 | Sonnet 5.5 | max | 70.9% | $0.28 | $19 | 1.4 h |
| 2 | Gemini 4 Carbon | high | 70.4% | $11 | $11 | 48 min |
| 3 | Opus 5.5 | max | 70.2% | $0.43 | $28 | 1.4 h |
| 4 | Gemini 4 Argon | high | 69.2% | $4.77 | $13 | 45 min |
| 5 | GPT-6 Astra | max | 65.0% | $1.82 | $21 | 34 min |
| 6 | GPT-6.1 Sol | max | 63.9% | $0.42 | $3.61 | 44 min |
| 7 | DeepSeek V4.1 Flash 1017 | max | 61.3% | free | $0.66 | 40 min |
| 8 | GPT-6.1 Terra | max | 60.5% | $2.92 | $2.92 | 40 min |
| 9 | GLM 5.3 | max | 59.7% | $0.89 | $8.52 | 1.1 h |
| 10 | Haiku 5.5 | max | 58.7% | $0.067 | $3.17 | 50 min |
| 11 | Grok 4.7 | xhigh | 58.6% | $1.08 | $12 | 34 min |
| 12 | MiMo V2.6 Pro | high | 57.8% | $0.36 | $0.43 | 1.2 h |
| 13 | Gemini 3.8 Flash | high | 56.0% | $0.20 | $5.03 | 36 min |
| 14 | MiMo V2.6 Flash | high | 54.8% | free | $0.22 | 1.1 h |
| 15 | DeepSeek V4 Pro 0813 | max | 53.0% | $2.50 | $2.50 | 58 min |
| 16 | Kimi K3 | max | 53.0% | $0.55 | $7.17 | 1.2 h |
| 17 | GPT-6 Luna | max | 52.3% | $0.067 | $0.45 | 32 min |
| 18 | GLM 5.3 Flash | max | 50.9% | free | $1.13 | 1.1 h |
| 19 | Mercury 3 | high | 43.7% | $0.48 | $0.48 | 22 min |
| 20 | Inkling | max | 29.0% | free | $1.85 | 17 min |
| 21 | Inkling Small | max | 27.6% | free | $0.40 | 16 min |