Claude speeds up 30 open biology models about 4x; Anthropic backs a design contest
See how this story connectsAnthropic reported that Claude optimised more than 30 open-source biomolecular models in under four weeks, for structure prediction, protein design, protein language models and genomics, with about 4x average speed-up at minimal precision loss and nearly 2x with identical outputs. Its FlashPairformer kernels beat NVIDIA’s cuEquivariance by 2.7 to 2.9x on triangle attention and 1.7 to 3.2x on multiplication. A low-memory mode folded systems over 10,000 tokens on one GPU node, including human mitochondrial complex I and a bacterial ribosome, while very large capsids were not predicted correctly. One Claude model with one H200 for 24 hours matched a protein design campaign that had used about 100 times more GPU hours. The code is open-sourced, and Anthropic co-sponsors a design competition with Adaptyv Bio, Modal and Twist Bioscience with wet-lab validation for more than 5,000 designs, up to $1m in Claude credits and $250,000 in Modal compute.
Why it matters
A lab is publishing measured speed-ups for the open biology stack and funding a wet-lab contest, the week it opened gated access to the same capabilities.
- When
- 17 Sep 2026
- ID
WTA-N-2026-0263- Where
- United States
- Who
- Anthropic, Adaptyv Bio, Modal, Twist Bioscience, NVIDIA
- Topics
- Biology, Drug discovery, Open-weight models, Research
Summary by wheresthe.ai, from Anthropic. Checked against the source on 12 Oct 2026.
Source copy: download the page as saved on 18 Oct 2026 (PDF). Document the source offers: document 1 (PDF). Also held by the Internet Archive.
Cite this story
WTA-N-2026-0263: “Claude speeds up 30 open biology models about 4x; Anthropic backs a design contest.” wheresthe.ai, 17 Sep 2026. https://wheresthe.ai/d/anthropic-biomolecular-modeling-uplift-2026-09/ (source: Anthropic).
Spotted an error? Email [email protected].