Claude Opus 4.8 has officially launched, maintaining the same pricing structure as version 4.7: $5 per million input tokens and $25 per million output tokens. This capability enhancement comes without a cost increase, a significant advantage for those operating at scale.
Key improvements include:
- Reduced Coding Errors: The new model is approximately four times less likely to overlook coding errors, enhancing reliability during extensive coding sessions.
- Increased Speed: Fast Mode operates at about 2.5 times the speed of the standard model, providing full Opus 4.8 functionality without compromising performance.
- Benchmark Gains: Notable increases in performance metrics include agentic coding rising from 64.3% to 69.2%, and multidisciplinary reasoning with tools improving from 54.7% to 57.9%. Additionally, a score of 84% was achieved on Online-Mind2Web for computer-use and browser-agent tasks.
- Dynamic Workflows: A research-preview feature in Claude Code allows for managing complex tasks across numerous parallel subagents, facilitating codebase-scale migrations across extensive lines of code.
