O'zapft is! 🍺 Suppa Lederhosn karts
Kotzhügel Fight (3 hearts), online rooms without codes & flight route in Stahlgraben 🚀 In the browser, no download. A joint work of Blender, Unreal, ElevenLabs, Opus 5, Opus 5.5, Astra 6 & GLM 5.3. Play:
Claude Sonnet 5.5 jumped from 10.3% to 70.6% on Terminal-Bench, a coding benchmark.
It nearly matches Opus 5.5 on knowledge-work benchmarks, and costs up to 30% less per task than Sonnet 5, per The Decoder. Same price per token. It just needs far fewer tokens for the same work.
Anthropic's cheaper model just beat Opus at agentic coding.
Claude Sonnet 5.5 is out: - 70.6% vs Opus 5.5's 66.4% on Terminal-Bench 4.0 - 30%+ faster than Sonnet 5 - Up to 30% cheaper for most work - Half Opus's price per token Top-tier AI is becoming everyday AI.
Claude Sonnet 5.5 is now available.
・More than 30% faster than Sonnet 5 ・Up to 30% cheaper per job (price list remains the same) ・Development evaluation 10.3% → 70.6% ・Input 2 dollars ・Output 10 dollars / 1 million tokens = half of Opus 5.5 Use Opus for difficult decisions and Sonnet for everyday tasks.
You have reached the end of the archive
All of Claude Opus 5