Claude Opus 4.8 Rolls Out: Anthropic Strikes Back in AI Race

Anthropic has activated Claude Opus 4.8 for users on May 28, 2026, just weeks after Opus 4.7’s April launch.

Fresh code leaks, desktop app sightings, and backend references confirm the rollout, delivering stronger agentic coding and reasoning amid intensifying competition from OpenAI.

Claude Opus 4.8 Launches: Anthropic Upgrades AI Flagship

Anthropic officially released Claude Opus 4.8 on May 28, 2026, delivering measurable improvements over Opus 4.7. The release confirms earlier speculation, after leaks on reddit suggested a planned roll-out.

Notwithstanding, the new model is now available at the same price with powerful new features for coding, agentic workflows, and user control.

Major Capability Gains

Opus 4.8 shows stronger performance across coding, agentic skills, reasoning, and practical knowledge work benchmarks.

Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors.Available today at the same price. pic.twitter.com/EufxL7T1kb

— Claude (@claudeai) May 28, 2026

Follow us on X to get the latest news as it happens

Early testers reportedly highlight greater reliability, sharper judgment, and significantly improved honesty. The model is four times less likely than Opus 4.7 to miss flaws in code it produces and is less prone to unsupported claims.

Alignment assessments also reached new highs in prosocial traits while showing substantially lower rates of misaligned behavior compared to Opus 4.7.

New features rolling out today include:

• Effort Control: Users on claude.ai and Cowork can now select how much thinking effort Claude applies — from Low (faster, lower rate-limit usage) to Max. Opus 4.8 defaults to High effort for the best balance of quality and experience.
• Dynamic Workflows in Claude Code: This research preview feature enables Claude to tackle massive tasks by planning, running hundreds of parallel subagents, and verifying outputs. It supports codebase-scale migrations across hundreds of thousands of lines of code.
• Messages API Update: Developers can now insert system instructions mid-conversation without breaking prompt cache.

Pricing and Availability

Standard pricing remains unchanged: $5 per million input tokens and $25 per million output tokens.

Fast Mode for Opus 4.8, running at 2.5× speed, is priced at $10/$50 and is three times cheaper than previous fast modes.

The model is available immediately across claude.ai, Claude API (claude-opus-4-8), and major cloud platforms.

Anthropic plans lower-cost models with similar capabilities and is preparing Mythos-class models for wider release in the coming weeks after completing stronger cyber safeguards under Project Glasswing.

Enterprises and developers can begin testing Opus 4.8 today on complex agentic and coding workloads.

Subscribe to our YouTube channel to watch leaders and journalists provide expert insights