Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort) — benchmark results
Claude Sonnet 4.6 evaluated in adaptive-reasoning mode at max effort. Provider: Anthropic. Released 2026-02-17. Access: API.
Unified ELO 1868 ± 20, rank #89 of 1776 rated models, from 59 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA Omniscience - Software Engineering (SWE) - HTML | 78 | Accuracy (%) | 96.2 |
| AA Terminal-Bench Hard | 53.03 | Accuracy (%) | 96.2 |
| AA Global-MMLU-Lite - Japanese | 92.58 | Accuracy (%) | 96.1 |
| AA Global-MMLU-Lite - Portuguese | 92.5 | Accuracy (%) | 96 |
| AA Omniscience - Software Engineering (SWE) - Rust | 80 | Accuracy (%) | 96 |
| AA Omniscience - Software Engineering (SWE) - Swift | 80 | Accuracy (%) | 95.9 |
| AA Global-MMLU-Lite - Bengali | 90.92 | Accuracy (%) | 95.8 |
| Tau3 Banking | 30.52 | Success Rate (%) | 95.8 |
| AA Global-MMLU-Lite - Italian | 92.92 | Accuracy (%) | 95.7 |
| Artificial Analysis Intelligence Index | 47.21 | Intelligence Index | 95.6 |
| UGI - Writing | 64.15 | Writing Score | 95.5 |
| AA Omniscience - Software Engineering (SWE) - C | 83 | Accuracy (%) | 95.1 |
Interactive version: theaggregate.ai/model?slug=claude-sonnet-4-6-adaptive-reasoning-max-effort · How the rankings work · Data refreshed daily, snapshot 2026-07-22.