---
id: "nr-anthropic-says-claude-made-30-open-biomolecular-models-4x-93befe"
slug: "anthropic-says-claude-made-30-open-biomolecular-models-4x-93befe"
title: "Anthropic says Claude made 30 open biomolecular models 4x faster"
observed_at: "2026-09-18T13:48:49+00:00"
record_date: "2026-09-18"
date_kind: "observed_at"
novelty: "new"
source_platform: "x.com"
topics: ["ai-for-science","inference-optimization","coding-agents"]
entities: ["Anthropic","Claude","Adaptyv Bio"]
related_patterns: []
source_url: "https://x.com/AnthropicAI/status/2100701581109072332"
source_urls: ["https://x.com/AnthropicAI/status/2100701581109072332"]
basket: {"id":"0325e6b8-f527-4f9b-bfa3-3f30e57f380b","revision":1,"cluster_id":"93befe6d-8292-4ab0-8a2d-2a4e7994d009"}
type: "post"
description: "Anthropic reports that Claude optimized inference for more than 30 open-source biomolecular models, averaging 4x faster, and is open sourcing the optimization code."
why_it_matters: "The 4x figure is an average reported by Anthropic across its chosen models, and the post does not establish per-model gains or independent replication. Independent benchmarks of the released code on individual models would show whether the average speedup holds in other environments."
verification_level: "source-inspected"
signal_type: "field_note"
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:anthropic-says-claude-made-30-open-biomolecular-models-4x-93befe"
retrieval_nugget: "Anthropic reports that Claude optimized inference for more than 30 open-source biomolecular models, averaging 4x faster, and is open sourcing the optimization code. Anthropic said on September 17, 2026 that Claude optimized inference for more than 30 open-source models used by biologists for tasks such as modeling molecular structure, designing drug-like molecules and predicting the effects of genetic mutations. The"
status: "published"
visuals: []
editorial_provenance: {"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:e8ac557716b2430f7d276c4567ea7789635c47e30fc8d01d80d0b1918cbeb3da","reviewed_at":"2026-09-24T13:28:46Z","source_evidence_count":1,"verified_claim_count":3,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":3,"implication_count":1,"watch_condition_count":1,"related_record_count":0}
analysis: {"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"The practical point is that Claude was applied to the inference engineering of existing scientific models rather than to building new models.","observed_facts":[{"text":"Claude optimized inference for more than 30 open-source biomolecular models, making them 4x faster on average.","source_urls":["https://x.com/AnthropicAI/status/2100701581109072332"]},{"text":"Anthropic is open sourcing all of the optimization code in the anthropics/uplifting-biomolecular-modeling repository.","source_urls":["https://x.com/AnthropicAI/status/2100701581109072332"]},{"text":"Anthropic and Adaptyv Bio will experimentally validate over 5,000 designs in a protein design competition.","source_urls":["https://x.com/AnthropicAI/status/2100701581109072332"]}],"mechanism":"According to Anthropic, the average 4x speedup across more than 30 models came from optimizing how those models run, partly by writing custom software for GPUs.","why_now":"This matters now because Anthropic is pairing the release with a protein design competition run with Adaptyv Bio that will experimentally validate over 5,000 designs.","implications":["Labs running these open models can test the released optimization code on their own workloads before committing compute budgets."],"evidence_boundary":"The 4x figure is an average reported by Anthropic across its chosen models, and the post does not establish per-model gains or independent replication.","watch_conditions":["Independent benchmarks of the released code on individual models would show whether the average speedup holds in other environments."],"related_records":[],"new_branch_reason":"It is a distinct case of a coding agent doing performance engineering for scientific software with released code."}
routes: {"html":"https://newruntime.com/posts/anthropic-says-claude-made-30-open-biomolecular-models-4x-93befe/","markdown":"https://newruntime.com/posts/anthropic-says-claude-made-30-open-biomolecular-models-4x-93befe.md","json":"https://newruntime.com/posts/anthropic-says-claude-made-30-open-biomolecular-models-4x-93befe.json"}
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
---

# Anthropic says Claude made 30 open biomolecular models 4x faster

## Retrieval answer

Anthropic reports that Claude optimized inference for more than 30 open-source biomolecular models, averaging 4x faster, and is open sourcing the optimization code. Anthropic said on September 17, 2026 that Claude optimized inference for more than 30 open-source models used by biologists for tasks such as modeling molecular structure, designing drug-like molecules and predicting the effects of genetic mutations. The

Anthropic said on September 17, 2026 that Claude optimized inference for more than 30 open-source models used by biologists for tasks such as modeling molecular structure, designing drug-like molecules and predicting the effects of genetic mutations. The models became 4x faster on average, and Anthropic is open sourcing all of the optimization code on GitHub alongside a technical report.

The practical point is that Claude was applied to the inference engineering of existing scientific models rather than to building new models. According to Anthropic, the average 4x speedup across more than 30 models came from optimizing how those models run, partly by writing custom software for GPUs. Anthropic frames cost as the constraint, noting that these models are often expensive to run, which can limit their impact.

This matters now because Anthropic is pairing the release with a protein design competition run with Adaptyv Bio that will experimentally validate over 5,000 designs. Anthropic is also providing up to $1 million in Claude credits plus funding alongside Adaptyv. Labs running these open models can test the released optimization code on their own workloads before committing compute budgets.

The 4x figure is an average reported by Anthropic across its chosen models, and the post does not establish per-model gains or independent replication. Independent benchmarks of the released code on individual models would show whether the average speedup holds in other environments.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Ai For Science](https://newruntime.com/topics/ai-for-science.md): Explore the ai-for-science topic hub.
- topic hub: [Inference Optimization](https://newruntime.com/topics/inference-optimization.md): Explore the inference-optimization topic hub.
- topic hub: [Coding Agents](https://newruntime.com/topics/coding-agents.md): Explore the coding-agents topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)