Anthropic · Language model

Claude Sonnet 5.5

Anthropic's mid-tier model, upgraded a week after Opus 5.5. Same price as Sonnet 5, and Anthropic says it is more than 30% faster and up to 30% cheaper per task because it uses fewer tokens and tool calls.

#5Power Rankings#2Coding#6Best Value
MakerAnthropic
ReleasedSeptember 28, 2026
API price (per 1M tokens)$2 in / $10 out
Open weightsNo

Best for

  • Everyday agentic coding
  • High-volume work where Opus is overkill
  • Fast drafting and analysis

Strengths

  • Beats Opus 5.5 on Terminal-Bench 4.0 in Anthropic's and Artificial Analysis's runs
  • Debuted in the Arena WebDev top four
  • 1M-token context window

Watch out for

  • Opus 5.5 still leads clearly on SWE-Bench Pro
  • Higher-risk cyber requests fall back to Sonnet 5

Ranking history

EditionPower RankingsCodingBest ValueOpen-Weight WatchImage GenerationVideo Generation
Week 45, 2026#5 (92)#2 (94)#6 (88)———
Week 44, 2026#3 (94)#2 (94)#6 (89)———
Week 43, 2026#3 (94)#2 (94)#5 (90)———
Week 42, 2026#2 (95)#2 (94)#4 (91)———
Week 41, 2026#2 (95)#2 (94)#3 (92)———

Head-to-head