Anthropic Releases Claude Opus 4.8 at Flat Pricing
Anthropic released Claude Opus 4.8 on 2026-05-28, improving agentic coding performance to 69.2% on SWE-bench (up from 64.3% in 4.7, beating GPT-5.5 at 58.6%). The model is 4× less likely to leave its own code flaws unremarked, adds a Dynamic Workflows research preview that spawns hundreds of parallel verifying sub-agents, and brings a 1M-token context window. Standard pricing stays at $5/$25 per million tokens; Fast mode drops to $10/$50 — 3× cheaper — at 2.5× speed, toggled via /fast in Claude Code.
Why It Matters
Opus 4.8 raises the agentic coding baseline at flat cost, making it directly relevant to any enterprise or developer running Claude-based workflows; the Dynamic Workflows feature in particular mirrors patterns that previously required custom orchestration infrastructure.