Claude Opus 4.8 Released: Better Reasoning, Cheaper Fast Mode
Opus 4.8 is out
Better reasoning, more efficient tool calling, 4x less likely to miss bugs in its code without comments
- Dynamic Workflows in Claude Code — hundreds of parallel sub-agents in a single session, migrations of hundreds of thousands of lines of code
- Effort Control on claude.ai — you choose how deeply the model thinks
- Fast mode is now 3x cheaper
84% on Online-Mind2Web (best browser-agent model), a record on Legal Agent Benchmark, ahead of GPT-5.5 on Super-Agent
$5 / $25 per 1M tokens (input/output)
They’ll soon release Mythos-class models — a higher level of intelligence than Opus, launching in the coming weeks