Anthropic has released Claude Opus 4.8, maintaining the same per-token pricing as its predecessor but increasing the default effort level for reasoning. This change means tasks will consume more output tokens, effectively raising costs for users without altering the per-token rate. The new version also introduces dynamic workflows for parallel subagent execution and claims improved honesty and benchmark performance, positioning itself as a cost-competitive alternative to models like GPT-5.5. AI
IMPACT Increases operational costs for developers using Claude Opus by defaulting to higher reasoning token usage.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on dev.to — Claude Code tag →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →