Anthropic AI Updates: April 30, 2026

1. Anthropic Releases BioMysteryBench to Evaluate Claude on Real Bioinformatics Research

Anthropic. Anthropic published BioMysteryBench, a benchmark of real-world bioinformatics research problems used to evaluate Claude’s scientific reasoning. Newer Claude generations match or exceed the performance of a human-expert panel on tasks the experts couldn’t solve independently, framing Claude as a credible research collaborator in computational biology rather than just a code-completion or literature-search tool. The benchmark complements Anthropic’s recent push into scientific applications and provides a harder, less gameable evaluation than synthetic biology question banks. Source

2. Anthropic in Talks to Raise ~$50B at an $850-900B Valuation

Anthropic. Multiple preemptive offers value Anthropic at $850-900B (up from $380B in February), with around $50B reportedly on the table. The round is driven by a revenue run-rate that has jumped from roughly $9B at the end of 2025 to north of $30B (and reportedly closer to $40B), largely on Claude Code adoption. The implied multiple still trails OpenAI’s most recent marks but narrows the gap meaningfully and signals that enterprise coding spend on Anthropic is now structural rather than experimental. Source