Anthropic releases Claude Opus 4.8 at the same price

Anthropic introduced Claude Opus 4.8 on 28 May 2026, describing it as an upgrade of Opus 4.7 rather than a new generation of the model. The company says the new version is available immediately at the same price as its predecessor and is meant to be a more effective collaborator on long, multi-step work.

What changed in the model

According to Anthropic, Opus 4.8 improves results on benchmarks covering coding, agentic skills, reasoning and practical knowledge-work tasks. A comparison table published with the announcement sets the model against Opus 4.7 and against rival systems, while a wider set of evaluations is provided in the accompanying Claude Opus 4.8 System Card. Pricing is unchanged: the model ships at the same cost as the version it replaces.

New controls and a faster mode

Three product changes arrive alongside the model. On claude.ai, users can now control how much effort Claude spends on a task, a setting that trades speed against thoroughness. Claude Code gains a dynamic workflows feature that Anthropic says lets it tackle very large-scale problems. Fast mode, in which the model works at 2.5 times its normal speed, is now three times cheaper than it was for previous models.

What early testers say

Anthropic published feedback from partners with early access. Staff engineer Tom Pritchard says the model asks the right questions, catches its own mistakes and pushes back when a plan is not sound. Kay Zhu, co-founder and CTO, reports that Opus 4.8 was the only model to complete every case end-to-end on his company's Super-Agent benchmark, beating earlier Opus models and GPT-5.5 at parity on cost. Michael Truell, co-founder and CEO of Cursor, says the model exceeds prior Opus versions at every effort level on CursorBench and that tool calling is more efficient, using fewer steps for the same level of intelligence. Niko Grupen says it recorded the highest score on a legal-agent benchmark and became the first model to break 10 percent overall on the all-pass standard. One tester reports 84 percent on Online-Mind2Web, a computer-use and browser-agent benchmark.

Why it matters

Opus 4.8 is a same-generation release at the same price, so Anthropic is not charging more for higher benchmark scores — a key argument in a market where each major lab competes on cost per task. The new settings shift the balance between speed, spending and depth to the user, and the published testimonials concentrate on reliability in agentic work rather than on raw capability.