Anthropic's Claude Opus 4.8 Launches, Sets New Legal AI Benchmarks

Anthropic has launched Claude Opus 4.8, its latest flagship large language model, on May 28, 2026. Designed for complex legal workflows, the model has set new performance benchmarks, achieving a record-breaking 10.4% score on Harvey’s Legal Agent Benchmark (LAB) and 91.1% on the BigLaw Bench.

These metrics represent significant improvements over its predecessor, Opus 4.7, which scored 7.1% on LAB. The LAB test evaluates end-to-end completion of intricate legal tasks under a stringent all-pass system. Meanwhile, the BigLaw Bench results show 43% perfect scores and 88% of results scoring above 0.80, with standout performance in deal management, compliance, and risk assessment.

According to Niko Grupen, Head of Applied Research at Harvey, Opus 4.8’s unique ability to self-review and refine its outputs has contributed to its superior drafting accuracy. "Opus 4.8 is the first model to break 10% on LAB’s tough all-pass scoring. Its review-and-revise behavior leads to stronger performance on drafting tasks," he stated.

The model also excels in legal accuracy across practice areas, including identifying correct case captions, parties, and statutory provisions while applying doctrinal frameworks accurately. However, evaluators noted some room for improvement in handling ambiguity and flagging areas requiring human review.

Beyond legal tasks, Opus 4.8 introduces enhanced coding capabilities through its “dynamic workflows” tool, which supports large-scale codebase migrations. It also features faster processing speeds—2.5× faster in fast mode compared to earlier iterations—at a reduced cost. These updates solidify Opus 4.8’s position not just in legal tech but as a versatile AI platform for broader enterprise applications.

Opus 4.8 will be available to eligible clients in the U.S. and the EU in the coming days, while availability in Australia is expected to follow. The model is accessible via claude.ai, Anthropic’s API platform, and major cloud providers.

Anthropic’s release of Opus 4.8 comes as the company prepares to expand access to its more advanced “Claude Mythos” series, aimed at cybersecurity and high-safeguard use cases. With Opus 4.8’s launch establishing new performance benchmarks, it signals Anthropic’s intent to remain a leader in the competitive AI space.