Anthropic Releases Claude Sonnet 5.5: 70.6% on Terminal-Bench 4.0 at the Same $2/$10 Price
70.6% on Terminal-Bench 4.0. Anthropic's Sonnet 5.5 closes the gap to Opus while cutting cost-per-task by 30%—same price, 30% faster.
Why it matters
Sonnet 5.5 delivers measurable capability gains (within 2 points of Opus on GDPval-AA) and faster output at unchanged per-token pricing, reducing operational cost through token efficiency. Practitioners can deploy immediately across major cloud platforms; the real win is cost-per-task improvement without paying more per token.
The key facts
7 to knowClaude Sonnet 5.5 scores 70.6% on Terminal-Bench 4.0
Within 2 points of Opus 5.5 on GDPval-AA benchmark
30%+ faster output generation vs Sonnet 5
Same $2/$10 per million token pricing as Sonnet 5
Cost-per-task falls by up to 30% due to reduced token consumption
Available via Claude API, AWS, Google Cloud, and Azure
Second model in Claude 5.5 family
The story so far
Earlier coverage of this storyline
- Anthropic releases Sonnet 5.5, which it calls a significantly cheaper, faster work partnerTechCrunch AI
- Introducing Claude Sonnet 5.5 on AWSAWS Machine Learning Blog
- Claude Sonnet 5.5Simon Willison
- This story
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: Anthropic has released Claude Sonnet 5.5, the second model in its Claude 5.5 family. It scores 70.6% on Terminal-Bench 4.0 and lands within 2 points of Opus 5.5 on GDPval-AA. It also generates output 30%+ faster than Sonnet 5 and keeps the same $2/$10 per million token price. Anthropic says cost…