Best general-purpose LLM: Vals Index ranking
- Claude Sonnet 5.5 leads
- 70.8%
- models measured
- 20
- lead over #2
- +0.7 pts
- median
- 58.2%
Leader on Vals Index · read October 20
As of October 20, 2026, Anthropic Claude Sonnet 5.5 leads general use with 70.8% on Vals Index, ahead of Anthropic Claude Opus 5.5 (70.2%) and Google Gemini 4 Argon (69.2%). This ranking is one public board, shown as published, and is not part of the Vibe Coding Index.
| Rank | Model | Vals Index | Coding evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | Claude Sonnet 5.5 Anthropic | 70.8% | 88.9 | $10.00 | |
| 2 | Claude Opus 5.5 Anthropic | 70.2% | 92.2 | $20.00 | |
| 3 | Gemini 4 Argon Google | 69.2% | 80.5 | $10.00 | |
| 4 | GPT-6 Astra OpenAI | 65.0% | 88.7 | $50.00 | |
| 5 | Muse Wonder Meta | 64.9% | 64.5 | $20.00 | |
| 6 | GPT-6.1 Sol OpenAI | 63.9% | 88.7 | $10.00 | |
| 7 | DeepSeek V4.1 Flash 1017 DeepSeek | 61.3% | 83.3 | $0.60 | |
| 8 | GLM 5.3 Z.ai | 59.7% | 73.1 | $1.40 | |
| 9 | Claude Haiku 5.5 Anthropic | 58.7% | 75.9 | $0.50 | |
| 10 | Grok 4.7 xAI | 58.6% | 75.0 | $6.00 | |
| 11 | MiMo-V2.6-Pro Xiaomi | 57.8% | 72.6 | $0.87 | |
| 12 | Gemini 3.8 Flash Google | 56.0% | 70.5 | $4.125 | |
| 13 | MiMo-V2.6-Flash Xiaomi | 54.8% | 61.9 | $0.28 | |
| 14 | Hy4 preview Tencent | 53.2% | 61.6 | $2.501 | |
| 15 | DeepSeek V4 Pro 0813 DeepSeek | 53.0% | 54.6 | $1.98 | |
| 16 | Kimi K3 Moonshot AI | 53.0% | 61.8 | $11.936 | |
| 17 | GPT-6 Luna OpenAI | 52.3% | 71.4 | $0.50 | |
| 18 | GLM 5.3 Flash Z.ai | 50.9% | 71.0 | $0.16 | |
| 19 | T Inkling Thinking Machines | 29.0% | 49.7 | $4.05 | |
| 20 | T Inkling Small Thinking Machines | 27.6% | 29.9 | $1.20 |
What this measures
How well does the model do professional tasks across several fields at once?
The Vals Index from vals.ai is a composite over many kinds of professional tasks. It measures general capability, not coding.
Read from Vals Index on October 20, 2026; the board says it was last updated October 19, 2026. Each model appears once with its best published setting; rows that match no model in our catalog are left out. Vals Index benchmark page