Claude Speeds Up Open-Source Biomolecular Models and Cuts Compute Costs

Claude Speeds Up Open-Source Biomolecular Models and Cuts Compute Costs Anthropic announced that Claude optimized more than 30 open-source biomolecular models, delivering an average 4x speedup, enabling single-node structural modeling of large molecular machines, and reducing the computational resources required for protein design by two orders of magnitude. Claude optimized over 30 open-source biological models across structure prediction, design, and genomics in under four weeks, achieving an average 4x speedup. Anthropic and Claude developed FlashPairformer, custom kernels outperforming field standards by 2.7–2.9x on triangle attention and 1.7–3.2x on triangle multiplication. A low-memory 'Big' mode enabled accurate structure prediction of complexes exceeding 10,000 tokens—and inference up to 70,000 tokens—on a single GPU node. Streamlined protein design workflows matched previous in silico binding prediction scores while reducing GPU hours by two orders of magnitude, bringing total spend down to approximately $150. Kernel Development and Model Acceleration Biomolecular structure prediction models like AlphaFold3 spend significant runtime and memory on triangle attention and triangle multiplication operations, which scale cubically. To address this bottleneck, Claude developed FlashPairformer, a suite of custom kernels that outperform the cuEquivariance standard by 2.7–2.9x on triangle attention and 1.7–3.2x on triangle multiplication. Working under the supervision of researchers with no kernel engineering background, Claude applied model-specific optimizations—such as caching recomputed operations and removing dead branches—across more than 30 open-source biology models in under four weeks, yielding an average 4x speedup with minimal impact on precision. Scaling to Large Molecular Complexes Predicting the structures of large multi-component cellular machinery typically requires distributed inference across multiple GPU nodes. Claude introduced a low-memory 'Big' mode that allows accurate modeling of complexes exceeding 10,000 tokens on a single NVIDIA GPU node. Successfully folded structures included human mitochondrial complex I, the TRiC chaperone complex, a proteasome, and a bacterial 70S ribosome. On an 8-GPU NVIDIA B300 node, the model executed inference on viral capsids and protein compartments reaching up to 70,000 tokens, significantly expanding single-node computational scale despite these sizes falling far outside the original training context. Efficient Protein Design Workflows Anthropic simplified its earlier agentic protein design pipeline, which previously used up to $10,000 in compute (around 2,500 NVIDIA H100 GPU hours) and extensive prompts. In new evaluations, a single Claude instance operating on one NVIDIA H200 for 24 hours with an 1,100-word prompt matched previous in silico ipSAE binding scores across 16 targets. Across evaluated Claude models, these design campaigns reduced GPU hours by roughly two orders of magnitude, matching prior performance for approximately $150 in combined GPU and token expenditures. Open-Source Release and Design Competition Anthropic open-sourced the optimized code repositories for the community and partnered with Adaptyv Bio, Modal, and Twist Bioscience to launch an open protein design competition. Focusing on five advanced challenges, the competition provides participants up to $1 million in Claude credits, $250,000 in Modal compute credits, and experimental wet lab validation for more than 5,000 submitted designs. Anthropic also opened public beta enrollment for its Life Sciences Verification Program to provide frontier AI access to biological researchers.

Anthropic 利用 Claude 大幅优化开源生物分子建模与蛋白质设计 Anthropic 宣布利用通用研究模型 Claude 在不到四周内对 30 多个开源生物模型完成推理与内存优化，平均提速约 4 倍，并将代码完全开源。结合新的推理核与精简架构，团队在单 GPU 节点上实现了超万 token 级分子系统的准确预测，同时使从头蛋白质设计的算力开销降低了约两个数量级。 Claude 在不到四周时间内优化了 30 多个生物学模型，平均提速约 4 倍（无损精度下近 2 倍），且开源了全部优化代码。 团队开发了名为 FlashPairformer 的自定义内核，在三角注意力和三角乘法运算上显著超越 NVIDIA cuEquivariance 标准。 低显存 Big 模式使单个 GPU 节点能够准确预测超过 10,000 个 token 的大型生物大分子机器（如人类线粒体复合体 I 和 70S 核糖体）。 简化后的单 Agent 流程在单张 H200 上运行 24 小时，仅花费约 150 美元即达到了此前花费 10,000 美元计算资源实现的计算结合评分（ipSAE）。 Anthropic 联合 Adaptyv Bio 发起蛋白质设计竞赛，提供高达 100 万美元的 Claude 额度并对超 5,000 个设计进行湿实验验证。 自定义内核与模型级推理提速 现代蛋白质结构预测模型（如 AlphaFold3、OpenFold3 和 Boltz-2）的计算瓶颈主要集中在时间与显存复杂度呈三次方增长的三角注意力和三角乘法操作上。在缺乏内核工程经验的研究人员监督下，Claude 开发了名为 FlashPairformer 的自定义内核，相比行业基准 NVIDIA cuEquivariance，三角注意力提速 2.7 至 2.9 倍，三角乘法提速 1.7 至 3.2 倍。此外，Claude 还针对具体模型实施了冗余缓存与死分支简化等优化，使 30 多个涵盖结构预测、基因组学和蛋白质语言模型的系统综合提速达 4 倍，且下游任务精度未受损。 单节点支持超大规模生物分子机器预测 针对以往需要多 GPU 节点才能预测的超大分子复合体，Claude 构建了低显存的“Big”模式。该模式在单台 NVIDIA GPU 节点上支持准确预测超过 10,000 个 token 的体系，并成功解析了人类线粒体复合体 I、TRiC 伴侣蛋白复合体、蛋白酶体和 70S 细菌核糖体，结果与实验测定结构高度一致。尽管在单个 8 卡 B300 节点上尝试运行 31,000 至 70,000 token 的病毒衣壳与蛋白隔室时由于超出模型训练上下文而未能得出正确预测，但该实验验证了单节点支持该量级推理的可行性。 以极低算力成本实现同等蛋白质设计效果 在早期研究中，Claude 针对单一靶点的从头结合蛋白设计需消耗多 Agent 架构及约 10,000 美元（约 2,500 个 H100 GPU 小时）的算力。通过优化模型和简化提示词，团队将环境缩减为单模型、单张 H200 显卡、无子 Agent 及无人工干预的配置。Mythos 5.1、Mythos 5 和 Opus 5 在 16 个靶点上评估的 ipSAE 计算结合预测评分与此前大规模耗时活动基本相当，GPU 与 Token 的综合成本降至约 150 美元，GPU 用时减少了两个数量级。 社区开源与蛋白质设计竞赛 Anthropic 已将所有优化代码在 GitHub 上开源，并发布了技术报告。同时，公司联合 Adaptyv Bio、Modal 和 Twist Bioscience 发起针对跨物种反应性、pH 敏感性、pMHC 特异性及 GPCR 等前沿难题的设计竞赛，承诺提供 100 万美元 Claude 额度、25 万美元 Modal 算力以及对超 5,000 个设计的湿实验验证。此外，Anthropic 面向生命科学领域的生命科学验证计划（Life Sciences Verification Program）也已开启公开测试。

*In this post, we share how Claude made the open-source models that scientists use to predict and design biomolecules faster and more memory-efficient. Claude, working within [Claude Science,](https://claude.com/product/claude-science) optimized more than 30 of these models in just under four weeks, speeding them up roughly 4x on average. It also created a low-memory mode that enables the accurate prediction of biomolecular systems larger than 10,000 tokens (amino acids, nucleotides, and atoms from small molecules and ions) on a single NVIDIA GPU node. We are open-sourcing all of the optimized code and announcing a protein design competition co-sponsored with Adaptyv Bio, backed by up to $1 million in Claude credits and wet lab validation for over 5,000 designs.* Recently, we [shared results](https://www.anthropic.com/research/Claude-accelerates-protein-design) demonstrating Claude’s abilities to design *de novo* protein binders through expert-level orchestration of open-source protein design and structure prediction models. *De novo* binders are small, computationally designed proteins that attach tightly to a specific target molecule to activate, block, or deliver something to it. Although this was an encouraging demonstration of AI’s scientific capabilities and an early step towards advancing drug discovery, it took more resources than would be available to the vast majority of protein designers. We allowed Claude to spend up to $10,000 per target on the AI infrastructure platform Modal, roughly equivalent to 2,500 NVIDIA H100 GPU hours. To make such research more accessible, we began to explore inference optimizations to run these models more efficiently. As an early result of these optimizations, [Claude Mythos 5.1](https://www.anthropic.com/claude-fable-and-mythos-5-1) accelerated seven open-source biology models, enabling them to run up to 2.5 times faster. Here, we present new results showing how an internal, general-purpose research model was able to optimize more than 30 deep learning models trained for a variety of biological tasks, such as structure prediction and protein design, as well as for genomics and protein language models. On average, Claude was able to speed up such tasks roughly 4x while sacrificing a minimal amount of precision, and nearly 2x with identical outputs. Claude also improved the memory utilization of these models, making it possible to predict biomolecular systems of unprecedented sizes. By combining these results with simplifications to our previous agentic protein design approach, we show that Claude can achieve comparable *in silico* performance to the results we previously reported using two orders of magnitude fewer GPU hours. Beyond protein design, these specialized biological models are widely used by molecular biologists, including for drug discovery and development. We are open-sourcing the optimized code for all of these models today ([here](https://github.com/anthropics/uplifting-biomolecular-modeling)) so that the broader community can make use of them. You can find more detail in our technical report ([here](https://www-cdn.anthropic.com/e96b5807039a88168733d9687afe41dfbbd5de13.pdf)). To further support the community, we are also co-sponsoring a protein design competition with Adaptyv Bio, which has pioneered [open protein design competitions](https://proteinbase.com/competitions). We’ve jointly selected five challenging problems at the frontier of today’s capabilities. Together with Adaptyv, and thanks to generous contributions from Modal and Twist Bioscience, we’re committing up to $1 million in Claude credits and $250,000 in Modal compute credits, as well as wet lab validation for over 5,000 designs. Find more information ([here](https://proteinbase.com/competitions/anthropic-adaptyv-2026)) and ([apply here](https://docs.google.com/forms/d/e/1FAIpQLSc0Hz1ZWYTt_wkn76ViVxDghmEhG_OeVEcj9YGHxLWqxF1kWw/viewform?usp=dialog)). Protein structure prediction is the problem of determining the three-dimensional structure of a protein from its sequence of amino acids alone. Protein design, meanwhile, is the process of creating a protein with a specific structure, function, or set of properties. Together, these computational tools allow scientists to interrogate key biomolecular processes, such as how cancers form, and to create useful molecules, such as drugs that could target these cancers. Modern structure prediction models, such as AlphaFold3, OpenFold3, and Boltz-2, spend much of their computational runtime and memory on two operations: triangle attention and triangle multiplication, which act on triplets of tokens. These operations make it possible to model the geometry of biomolecular systems, but they are extremely computationally expensive, because they are cubic in both runtime and memory: doubling the size of the system uses 8x more time and memory, while tripling it uses 27x more. Writing kernels—low-level software translation layers for accelerated computing hardware such as GPUs—is a standard approach for reducing these costs. Given their significance, triangle attention and multiplication have been the subject of dedicated kernel development efforts, first with NVIDIA’s [cuEquivariance](https://github.com/nvidia/cuequivariance) and more recently with NVIDIA’s [BioNeMo Inference Runtime](https://github.com/NVIDIA-BioNeMo/BioNeMo-Inference-Runtime) (BioNeMo-IR). For our own effort to optimize inference for structure prediction models, we worked with Claude to develop FlashPairformer, a set of custom kernels that speed up triangle attention and multiplication. It achieves a new state-of-the-art, outperforming [the field standard](https://github.com/nvidia/cuequivariance) on average by 2.7-2.9x on triangle attention and 1.7-3.2x on triangle multiplication, depending on the model configuration. In addition to developing transferable kernels, we pointed Claude at each individual model with the goal of producing more specific optimizations. These included changes like caching redundant recomputed work and simplifying dead branches into their constant outputs. The combination of these improvements accelerated the structure prediction models by 4x, on average, and for each model, we confirmed that Claude’s accelerated versions did not impact performance on the downstream task (such as structure prediction). It normally takes an experienced team of engineers weeks to produce such optimizations for each model, and the work often does not transfer between models. Claude, supervised by two members of Anthropic’s technical staff who are experienced in biomolecular modeling but who had no prior experience in inference optimization or kernel engineering, carried out the acceleration of more than 30 open-source models across biomolecular structure prediction, protein design, protein language modeling, and genomics in just under four weeks. Our results suggest that frontier AI models will help others in the field build scientific tools with greater speed and ease. In addition to making these protein structure prediction and design models faster, we also tasked Claude with reducing the memory usage involved in modeling large molecular machines. Much of the work in a cell is done by such systems, including the ribosome that builds proteins, the respiratory complexes that power the cell, and the chaperones that help other proteins fold. Each is built from dozens of components, and its function depends on how those components fit together and interact. Predicting the structures of systems this large has typically required substantial computing resources inaccessible to most molecular biologists, such as inference spread across multiple GPU nodes. Claude created a low-memory “Big” mode that enables the accurate modeling of systems larger than 10,000 tokens and successful inference on systems larger than 70,000 tokens using just one NVIDIA GPU node—a previously out-of-reach task. Molecular machines folded successfully using Big mode include human mitochondrial complex I, the TRiC chaperone complex, a proteasome, and a bacterial ribosome, each closely matching its experimentally determined structure. To our knowledge, these are among the largest structures ever folded accurately using structure prediction models, with complex I and the 70S ribosome consisting of more than 10,000 tokens each, in comparison to the 40S ribosome predicted accurately by [AlphaFold3](https://www.nature.com/articles/s41586-024-07487-w), which consisted of 7,663 tokens. To test the limits of Claude’s optimizations, we asked Claude to predict structures of a greater size than anything that had previously been achieved. Using a single 8-GPU B300 node, Claude generated predictions of entire viral capsids and protein compartments ranging in size from more than 31,000 to more than 70,000 tokens. These systems are nearly two orders of magnitude larger than the training context of these structure prediction models, and, perhaps unsurprisingly, are not predicted correctly. However, the barrier to inferencing at this scale has been significantly lowered now that it takes just one NVIDIA B300 node, suggesting that with improved tools researchers will soon be able to computationally model an increasingly complex set of biological systems. In our earlier work on protein design, we provided Claude with an approximately 16,000-word prompt that encouraged it to utilize sub-agents and spend up to $10,000 per target on Modal (roughly 2,500 NVIDIA H100 GPU hours) in a 24-hour span. Here, we gave a single Claude model access to one NVIDIA H200 and 24 hours of wall time, a prompt of about 1,100 words, and a reference sheet for the pre-installed tools, with no sub-agents and no human steering the designs. We ran three Claude models (Mythos 5.1, Mythos 5, and Opus 5) against 16 targets with the accelerated biomolecular models described in this post. We scored designs by [ipSAE](https://www.biorxiv.org/content/10.1101/2025.02.10.637595v2), an *in silico* score that has been shown to be predictive of binding in the wet lab. Averaged over 16 targets, the median-scoring and highest-scoring designs from all three Claude models evaluated achieve approximately the same ipSAE values as our earlier Mythos 5.1 campaigns despite using about two orders of magnitude fewer GPU hours. We also considered Claude token costs and found that with a combined spend of approximately $150 on GPUs and tokens, we can achieve *in silico* performance matching the levels of our previous campaigns. The optimizations described above help us predict and design molecules more efficiently, while unlocking capabilities that would have otherwise been resource-prohibitive. To demonstrate the uplift they provide and the impact of Claude on molecule design more broadly, we’re partnering with Adaptyv Bio to launch [a protein design competition](https://proteinbase.com/competitions/anthropic-adaptyv-2026). We’ve selected five problems at the frontier of today’s protein design capabilities, including challenges such as species cross-reactivity, pH-sensitivity, and peptide-MHC specificity, as well as difficult targets such as GPCRs. With the Adaptyv team, we’ll be experimentally validating over 5,000 designs submitted by the community against these problems. We will be providing up to $1 million in Claude credits and additional funds for experimental validation at Adaptyv for participating researchers, Modal will provide up to $250,000 in compute credits, and Twist Bioscience will provide DNA for the competition. You can find more information, including eligibility criteria ([here](https://proteinbase.com/competitions/anthropic-adaptyv-2026)) and ([apply here](https://docs.google.com/forms/d/e/1FAIpQLSc0Hz1ZWYTt_wkn76ViVxDghmEhG_OeVEcj9YGHxLWqxF1kWw/viewform?usp=dialog)). We have also begun to provide frontier AI capabilities to life scientists for biology-related work via our Life Sciences Verification Program. We recently enrolled our first group of organizations, and opened up the program in public beta today. You can find more information ([here](https://www.anthropic.com/news/life-sciences-verification-program)). The following resources provide further technical depth and more detailed information about the results described above: