Announcements
New Model Release: Claude Opus 4.8
Claude Opus 4.8 reasoning model was officially launched on May 28, 2026.
Key Upgrades
-
Coding & Agent: 69.2% SWE-Bench Pro, 1890 Elo GDPval-AA (leads GPT-5.5)
-
Honesty: 75% fewer code defects hidden; less unfounded claims
-
Effort Control: 4 modes (Fast→Max); Fast = 2.5× speed, 1/3 cost
-
Dynamic Workflows: Orchestrate 100s of parallel subagents
-
Same Pricing: $5/$25 per million input/output tokens