System Card← leaderboard
Anthropic

Claude Sonnet

#8of 28 · general

Sonnet 5.5

1Mcontext
$2 / $10per 1M in / out
2026-09-28released
2026-06-01knowledge cutoff
50100Coding 85 · #3 of 28Agents 78 · #6 of 29Cyber 13 · #17 of 24Engineering 89 · #2 of 26Math 79 · #5 of 26Health 91 · #1 of 24Legal 55 · #9 of 27Professional 66 · #7 of 28Writing 46 · #9 of 21 · one sourceInstruct 54 · #11 of 25Hallucination 57 · #5 of 26Multimodal 60 · #7 of 24
Coding85#3
Agents78#6
Cyber13#17
Engineering89#2
Math79#5
Health91#1
Legal55#9
Professional66#7
Writing46#9
Instruct54#11
Hallucination57#5
Multimodal60#7

History

LiveBench score7274767880Apr 2026Jul 2026Oct 2026Claude Sonnet 4.6 · $3/M in · 2026-02-17 · still served · LiveBench 73.0 LiveBench score (0–100)Claude Sonnet 5 · $3/M in · 2026-06-30 · still served · LiveBench 76.0 LiveBench score (0–100)Claude Sonnet 5.5 · $2/M in · 2026-09-28 · still served · LiveBench 77.8 LiveBench score (0–100)54.6
  • Anthropic
  • older → newer
$ / M input tokens3 of 176 priced models · price is a third-party claim, not a vendor quote

Effort

Sonnet 5.5
% of best score4060801000.2×0.5×1×2×× the best effort’s cost · log scaleLowMedHighXhighMax

Measured

≈ imputed · amber dot = lab-self-reported among the sources
GPQA Diamond95.6%#3Epoch AI
HLE50.5%#9Artificial AnalysisEpoch AI
LiveBench77.8%#9LiveBench
SciCode61.0%#5Artificial AnalysisEpoch AI
CritPt31.4%#4Artificial AnalysisEpoch AI
Terminal-Bench83.2%#4Vals.ai
APEX Agents75.5%#2Epoch AI
GDPval66.9%#3Artificial Analysis
Arena1476#11Arena
AA Index56.0#2Artificial Analysis
Vals Index67.0#3Vals.ai
Epoch Capabilities Index166.2#6Epoch AI

Neighbors

Kimi→ stronger codingGPT Sol→ near-twinGPT Astra→ more factual · pricierClaude Opus→ stronger professional · pricierClaude Fable→ pricier