Anthropic announced Claude Opus 4.8 on May 28, 2026, describing it as an upgraded version of Opus 4.7 with better benchmark performance, stronger collaboration, and improved reliability. The model is available immediately at the same standard price as its predecessor: $5 per million input tokens and $25 per million output tokens. A faster “fast mode” is also available at 2.5× speed, and it is now three times cheaper than for previous models.
The release comes with several new features. On claude.ai, users can now control how much effort Claude uses on a task, trading off speed, cost, and quality. Claude Code adds “dynamic workflows,” a research-preview feature that lets the model handle very large-scale tasks by planning work, running hundreds of parallel subagents, and verifying results before responding. The Messages API also now supports system entries inside the messages array, allowing developers to update instructions mid-task without disrupting prompt caching.
Anthropic says Opus 4.8 improves agentic performance, coding, reasoning, and practical work tasks. Early testers reported better judgment, more efficient tool use, stronger honesty, and fewer unsupported claims. The company also says the model shows lower rates of misaligned behavior and stronger prosocial traits. Anthropic says Opus 4.8 is a modest but meaningful step forward, while it continues developing lower-cost models and even more capable future systems.
Article timeline
Build this article’s timeline
Extract dated events, key actors, and open questions using the same timeline analysis available in Signal.
Anthropic announced Claude Opus 4.8 on May 28, 2026, describing it as an upgraded version of Opus 4.7 with better benchmark performance, stronger collaboration, and improved reliability.
🤖 AI prefill
📅WhenWhen did this happen, or what is the timeframe?
📍Where
Milan and Europe
🤖 AI prefill
💡Why
Centers on Claude Opus 4.8, dynamic workflows, and effort control
🤖 AI prefill
Login to answer and earn XP for each gap you fill.
Model UpgradeClaude Opus 4.8, upgrade, benchmarks, performance, priceAgentic Workflow Featuresdynamic workflows, Claude Code, effort control, parallel subagents, task planningEnterprise Use Casescoding, legal, research, browser agent, data workSafety Alignmenthonesty, alignment, misaligned behavior, deception, user autonomyAvailability Pricingavailable today, pricing, API, input tokens, output tokens
Key Phrases:Claude Opus 4.8100%dynamic workflows96%effort control90%agentic tasks88%computer-use and browser-agent model86%alignment assessment84%honesty82%tool calling80%enterprise AI78%Claude Code77%
Comments
Login to view and post comments.