# AlphaFold 3

AlphaFold 3。代码：Apache-2.0；权重：有限制。最后核实 2026-11-14。可经 MCP 或 API 直接查询。

## 代码许可

宽松 · Apache-2.0
商用：可以 · 改编：可以 · 再分发：可以

> Apache License
https://raw.githubusercontent.com/google-deepmind/alphafold3/main/LICENSE (2026-11-13)

## 权重许可

有限制 · LicenseRef-Google-DeepMind-terms
商用：不可以 · 改编：未写明 · 再分发：不可以

> The AlphaFold 3 model parameters and output are **only** available for
https://raw.githubusercontent.com/google-deepmind/alphafold3/main/WEIGHTS_TERMS_OF_USE.md (2026-11-13)

> You ***must not* publish or share AlphaFold 3 model parameters**
https://raw.githubusercontent.com/google-deepmind/alphafold3/main/WEIGHTS_TERMS_OF_USE.md (2026-11-13)

> The AlphaFold 3 model parameters and output are **only** available for non-commercial use by, or on behalf of, non-commercial organizations (*i.e.*, universities, non-profit organizations and research institutes, educational, journalism and government bodies).
https://raw.githubusercontent.com/google-deepmind/alphafold3/main/WEIGHTS_TERMS_OF_USE.md (2026-11-13)

> not (i) resident of a embargoed country, (ii) ordinarily resident in a US
https://raw.githubusercontent.com/google-deepmind/alphafold3/main/WEIGHTS_TERMS_OF_USE.md (2026-11-13)

> Google may at any time suspend or terminate your right to use and, as applicable
https://raw.githubusercontent.com/google-deepmind/alphafold3/main/WEIGHTS_TERMS_OF_USE.md (2026-11-13)

## AlphaFold 3 是否允许商业使用

- 代码：可以
- 权重：不可以
代码与权重的许可分别成立，一方的结论不外推到另一方。
结论仅记录来源所述，不构成法律意见。

## 对输出的限制

- 不得使用输出结果训练或构建与 AlphaFold 3 类似的生物分子结构预测机器学习模型或相关技术。
  > Use Output to train or create machine learning models or related technology for biomolecular structure prediction similar to AlphaFold 3
  https://raw.githubusercontent.com/google-deepmind/alphafold3/main/WEIGHTS_PROHIBITED_USE_POLICY.md (2026-11-13)
- 模型参数及其输出仅限大学、非营利组织、研究机构、教育机构、新闻机构和政府机构等非商业组织自身或代表其进行非商业使用。
  > The AlphaFold 3 model parameters and output are **only** available for non-commercial use by, or on behalf of, non-commercial organizations (*i.e.*, universities, non-profit organizations and research institutes, educational, journalism and government bodies).
  https://raw.githubusercontent.com/google-deepmind/alphafold3/main/WEIGHTS_TERMS_OF_USE.md (2026-11-13)
- AlphaFold 3 及其输出仅用于理论建模，并非为临床用途设计、验证或批准。
  > AlphaFold 3 and its output are for theoretical modeling only. They are not intended, validated, or approved for clinical use.
  https://raw.githubusercontent.com/google-deepmind/alphafold3/HEAD/README.md (2026-11-13)
- 《输出使用条款》（最后修改 2024-11-09）禁止代表商业机构或在商业活动中使用 AF3 输出，包括为其开展的研究。
  > Output are made available free of charge, for non-commercial use
  https://github.com/google-deepmind/alphafold3/blob/main/OUTPUT_TERMS_OF_USE.md (2026-11-13)
- 不得用 AF3 输出训练或构建与 AF3 类似的生物分子结构预测模型，包括通过蒸馏。
  > biomolecular structure prediction similar to AlphaFold 3 as made available
  https://github.com/google-deepmind/alphafold3/blob/main/OUTPUT_TERMS_OF_USE.md (2026-11-13)
- 来源原句：You must not share Output with any commercial organization.
  > You must not share Output with any commercial organization.
  https://raw.githubusercontent.com/google-deepmind/alphafold3/main/OUTPUT_TERMS_OF_USE.md (2026-11-13)
- 来源原句：Distribute Output, or disclose findings arising from using AlphaFold 3 without citing our paper:
  > Distribute Output, or disclose findings arising from using AlphaFold 3 without citing our paper:
  https://raw.githubusercontent.com/google-deepmind/alphafold3/main/WEIGHTS_PROHIBITED_USE_POLICY.md (2026-11-13)
- 来源原句：making decisions in domains that affect material or individual rights or
  > making decisions in domains that affect material or individual rights or
  https://raw.githubusercontent.com/google-deepmind/alphafold3/main/OUTPUT_TERMS_OF_USE.md (2026-11-13)

## 训练数据

- 训练所用结构数据均不晚于 2021-09-30 发布。
  > No structural data used during training were released after 30 September 2021
  https://pmc.ncbi.nlm.nih.gov/articles/PMC11168924/ (2026-11-13)
- 训练数据中加入了 AlphaFold-Multimer v2.3 的预测结构（交叉蒸馏）。
  > we use a cross-distillation method in which we enrich the training data with structures predicted by AlphaFold-Multimer (v.2.3)
  https://pmc.ncbi.nlm.nih.gov/articles/PMC11168924/ (2026-11-13)

## 运行方式

- 权重下载：https://storage.googleapis.com/alphafold3/af3.bin.zst
- 托管接口：https://alphafoldserver.com
- 第三方托管：https://github.com/google-deepmind/alphafold3/blob/main/README.md

## 各版本

- unspecified: code —, weights —
- AlphaFold 3: code —, weights —

## 尚无有出处的说明

已知缺陷与常见误用

https://mdexa.com/zh/models/alphafold3/

## 在本库继续查找

本条目是 109 条已处理并核实的记录之一（数据集、数据资源与开源模型）。查找其他条目：调用 `https://mcp.mdexa.com/v1/datasets?q=<关键词>`，或连接 MCP 服务 `https://mcp.mdexa.com/mcp`；每分钟 20 次以内无需密钥。每行返回不到 1 KB；空结果即"数据库未收录"。核对一份清单：一次批量核对（免费账号）即返回报告。说明：/agents.md。相似条目：Boltz (Boltz-1 and Boltz-2) (https://mdexa.com/zh/models/boltz/)；OpenFold3 (https://mdexa.com/zh/models/openfold3/)；ESM C (https://mdexa.com/zh/models/esm-c/)。