Anthropic launched Claude Opus 4.8 on May 28, 2026 — 42 days after Opus 4.7, which was released April 16, 2026. The release is positioned as a significant quality-of-life update: faster responses, better long-context consistency, and improvements across agentic and coding benchmarks.
Benchmark improvements
Anthropic highlighted gains on Terminal-Bench 2.1, OSWorld-Verified, Finance Agent v2, and Online-Mind2Web, alongside improvements to coding reliability. Opus 4.8 is also roughly four times less likely to let flaws in its own code pass without flagging them, and proactively signals uncertainty rather than confidently hallucinating.
Dynamic Workflows in Claude Code
The biggest agentic feature is Dynamic Workflows, a research preview that lets Claude Code plan a large task, run hundreds of parallel subagents in a single session, distribute the work across them, and verify outputs against your test suite — all without human intervention. Dynamic Workflows is available for Claude Code on Enterprise, Team, and Max plans, and is designed for repository-scale migrations and hundreds-of-thousands-of-lines-of-code refactors that would previously require a team of engineers.
Pricing: same base, cheaper Fast mode
Standard pricing holds at $5 per million input tokens and $25 per million output tokens — identical to Opus 4.7. The big change is in Fast mode: it drops from $30/$150 per million tokens (Opus 4.7) to $10/$50 — a 3× cost reduction for latency-sensitive workloads. Prompt caching brings savings up to 90%; batch processing offers 50% off.
Opus 4.8 is available in Claude.ai for Pro, Max, Team, and Enterprise subscribers, and via the Anthropic API, AWS Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.
Mythos Preview expanding access
Separately, Anthropic said it expects to bring Mythos-class models — its restricted-access cybersecurity models — to all its customers "in the coming weeks." Mythos Preview has already been used by a small number of organizations: during its 30-day preview, Mythos and roughly 50 partner organizations identified more than 10,000 high- or critical-severity vulnerabilities in systemically important software.