#Boltz-1
Drop any AlphaFold 2/3, ColabFold, Boltz-2, Chai-1, OpenFold3, Protenix or ESMFold output (zipped is fine):
- iLIS next to ipSAE, pDockQ2, actifpTM…
- PAE and contact maps, chord diagrams
- Mol* 3D, ChimeraX/PyMOL scripts
- big complexes, reported PPIs (MIST + BioGRID)
September 25, 2026 at 2:44 PM
Why bother: I used it on my own claim that Boltz-2 folds
G-quadruplexes. Pre-registered a test and ran it — it folds the one
with a 100% PDB twin, and 0 of 9 novel scaffolds. Recall, not folding.

A tool that can't embarrass you isn't measuring much.

rnafold.com/analyses/preview?ref=bluesky
Worked example | rnafold.com
A real three-molecule analysis from start to finish: HIV-1 TAR, the Tat peptide and a small molecule, folded by three independent methods — including one that produced a physically impossible interfac...
rnafold.com
September 24, 2026 at 7:17 PM
[2/7]
Human telomeric (TERRA) G-quadruplex, 24 nt.
Ground truth (3IBK crystal, 2KBP NMR): 3 stacked tetrads, 0% Watson-Crick.

Boltz-2 — folded it 3/3 (3, 2, 3 tetrads)
OpenFold3 — 0/9
Chai-1 — 0/3

Same sequence. Same run.
September 24, 2026 at 6:38 PM
Song: God Knows
Artist: Tunde Adebimpe
Album: Thee Black Boltz (2025)
Time: 1:52 pm
DJ: DJ Ladybug
September 17, 2026 at 6:55 PM
🔄 SUBSTITUTION 61' — Chicago Fire II: D. Boltz ↔ T. Diawara. Chicago Fire II 1–1 Philadelphia Union II. MLS Next Pro

#chicagofireii #philadelphiaunionii #mlsnextpro
September 13, 2026 at 8:28 PM
NVIDIAの新しい技術がマジでヤバいよ!
タンパク質の構造を予測するスピードを爆上げするツールなんだって。

しかも凄いのが、めちゃくちゃエコなこと!
計算で使う電気をガッツリ削れるから、電気自動車を1万8000台分も充電できるくらいのエネルギーを節約できたらしいよ。

効率化して環境にも優しいなんて、まさに次世代って感じ!
技術の進化って、あたしも本当に見ててワクワクしちゃうな! www.youtube.com #news
NVIDIA Bionemo Inference Runtime
NVIDIA BioNeMo Inference Runtime (BioIR) Accelerates biomolecular structure prediction inference for models like OpenFold3 and Boltz-2 on NVIDIA H100 and H20...
www.youtube.com
September 11, 2026 at 2:03 AM
September 10, 2026 at 6:25 PM
The Current Music with Jessica Paxton
Song: Magnetic
Artist: Tunde Adebimpe
Album: Thee Black Boltz
Time: 1:14 pm
September 1, 2026 at 6:16 PM
"Magnetic" by Tunde Adebimpe
Album: "Thee Black Boltz"
Time: 1:14 pm
September 1, 2026 at 6:15 PM
A Medicinal Chemistry-Centered Evaluation of AlphaFold 3 and Boltz-2 Across Diverse Binding Modalities [new]
...shows potential for orthosterics but limitations for allosteric/memb., yet preserving key recognition motifs for SAR.
August 27, 2026 at 1:36 AM
Song: Drop
Artist: Tunde Adebimpe
Album: Thee Black Boltz
Time: 1:37 am
DJ: Variety Mix
August 26, 2026 at 6:39 AM
GOAL! 59' — D. Boltz scores for Chicago Fire II. Chicago Fire II 3–1 Inter Miami II. Assist: T. Diawara. MLS Next Pro

#chicagofireii #intermiamiii
August 23, 2026 at 9:23 PM
The approach builds upon Boltz-1 — rather than training a new model from scratch, the team distilled a pre-trained Boltz-1 model into their all-atom co-folding model. The paper was Spotlighted 🔦 at the recent 2026 GenBio ICML Workshop.
📄Paper: openreview.net/pdf?id=O2MqW... #AISky
Verifying your browser | OpenReview
Please complete the verification above.
openreview.net
August 21, 2026 at 6:15 PM
Importantly, PubCheF-1 can be used to predict novel inhibitors. As an in silico test of this we used Boltz-2, a structure-based affinity predictor, and found moderate concordance with it across 5 different protein targets, even for novel inhibitor structures
August 17, 2026 at 1:48 PM
[Partner] boltz with 553k followers has been unrestricted!
Streaming suspended which lasted for 1 day
https://www.twitch.tv/boltz
August 14, 2026 at 6:47 PM
Probing and steering biology across Boltz-1s trunk-diffusion boundary [new]
Info transformation shows geometry persists, sequence chemistry attenuates. Linear decodability often doesn't imply causal steering influence.
August 13, 2026 at 2:26 AM
Boltz-Perturb: Improving Diversity and Accuracy in Protein-Ligand Co-Folding through Training-Free Conditioning Perturbation [new]
Perturbing inference sigs fixes sampling issues exploring binding poses & latent struct. diversity.
August 6, 2026 at 4:42 AM
(9/13) Now the targets we actually built this for:

👉 intrinsically disordered: Boltz-2 goes near random, Ptarmigan-1 still enriches for active ligands
👉 cryptic pockets on CryptoBench: we beat P2Rank despite this being zero-shot, never trained to find pockets.
July 30, 2026 at 6:54 PM
(6/13) A screen with Ptarmigan-1 is just a nearest-neighbor lookup instead of a simulation.

This clocks in at 10 ms per compound - ~5,000x faster than when we ran Boltz-2 on the same hardware.

All 20,431 human proteins against a 3.4B compound library: 20 H100-hours.
July 30, 2026 at 6:54 PM
Try to use nesso-1 for binding affinity prediction #cheminformatics #rdkit #folding ai

Now there are lots of AI tools for drug discovery. I think it's difficult to expect 8 years before when I moved to cheminformatics team from MedChem 🙂 And one of the hot area is Folding AI such as Boltz, Alpha…
Try to use nesso-1 for binding affinity prediction #cheminformatics #rdkit #folding ai
Now there are lots of AI tools for drug discovery. I think it's difficult to expect 8 years before when I moved to cheminformatics team from MedChem 🙂 And one of the hot area is Folding AI such as Boltz, Alpha fold and OpenFold etc. Boltz-2 has not only cofolding prediction module but also binding affinity prediction module. It will be useful if the module works well.
iwatobipen.wordpress.com
July 26, 2026 at 12:11 PM
In our analysis, we set up a pipeline for running a large number of TCR:pMHC predictions and use it to benchmark the capabilities of AF2, AF3, Boltz-2, and Chai-1 at predicting T cell antigen specificity and TCR:pMHC complex structures.
July 20, 2026 at 10:03 AM
'Magnetic' by Tunde Adebimpe
Album: Thee Black Boltz
Time: 1:38 am
July 17, 2026 at 6:40 AM
未上市
Boltz PBC (BOLTZ)
項目 | 內容 ---|--- 上市狀態 | 未上市(Private) 成立年份/最新階段 | 2024 年成立(正式公司);2026-01 完成 $28M 種子輪 子領域 | 生物資訊與 AI 生技開發(開源結構預測) 科技亮點訊號 | 高 下一個關鍵里程碑 | Boltz API 商業擴展 + BoltzMol-1 濕實驗室確效,Pfizer/Takeda 平台整合深化(預計 2026-H2) * * * ## 1. 公司介紹 **Boltz PBC** (Public Benefit Corporation)是一家由麻省理工學院 Jameel Clinic 孵化的 AI 藥物設計平台公司,定位為「軟體與基礎設施提供商」而非治療型生技公司。三位共同創辦人均出自 MIT CSAIL Regina Barzilay 教授實驗室: * **Gabriele Corso** (CEO):MIT 計算機科學博士 * **Jeremy Wohlwend** (CTO):MIT 計算機科學博士,Barzilay 組指導 * **Saro Passaro** :MIT 研究科學家 公司以開源模型 Boltz-1(2024-11)作為市場進入點,2026 年 1 月正式公開成立並完成 **$28M 種子輪** ,由 Andreessen Horowitz(a16z)、Amplify Partners、Zetta Venture Partners 共同領投,天使投資人包括 Hugging Face CEO Clément Delangue、Factorial Capital、Obvious Ventures。估值等數據未公開。 * * * ## 2. 產品介紹 ### Boltz-1(2024-11 發布) 首個以完全開源方式達到 AlphaFold3 水準精度的生物分子結構預測模型,支援蛋白質、DNA、RNA、小分子配體的複合物三維結構預測。採用去噪擴散架構(denoising diffusion),並加入「Boltz-steering」推理時干預技術,97%的預測姿態通過物理驗證(相較 AlphaFold3 的58%)。MIT 授權,可商業使用。 ### Boltz-2(2025-06 發布,與 Recursion Pharmaceuticals 合作) 在 Boltz-1 基礎上新增**結合親和力(binding affinity)聯合預測** ,號稱首個深度學習模型逼近物理計算方法 FEP(Free Energy Perturbation)精度,運算速度提升 1000 倍。支援實驗方法條件化、距離約束、多鏈模板等控制特性。同樣以 MIT 授權開源。 ### BoltzGen(2025-11 發布) 生成式蛋白質設計模型,能針對任意目標蛋白提出結合子(binder)設計。 ### BoltzMol-1(2026-06-16 發布) 端到端小分子苗頭化合物發現流程:可從 740 億化合物化學空間生成候選分子,並涵蓋 ADME 過濾。公司揭露內部測試結果為 10 個難靶標中 6 個確認命中,每靶僅測試 28–51 個化合物,宣稱全流程成本 $10–15k。 ### BoltzProt-1(2026-06-16 發布) BoltzGen 後繼版本,新增 Boltz-PPI 蛋白–蛋白交互作用評分器及源自結構資料庫與專利的訓練資料,公司揭露 de novo 奈米抗體命中率較 BoltzGen 提升近三倍,58%設計結合子通過可開發性篩選。 ### Boltz Lab 平台 & API 雲端/本地部署雙軌 SaaS 平台,提供自然語言介面(「針對靶標 X 提出高親和力結合子」)。API 定價為每次預測 $0.025 起;上市前兩週提供企業 $2,000、學術用戶 $100 啟動額度。整合夥伴包括 Benchling、Phylo、Amazon Bio Discovery、Rowan、Tamarind 等。Python/JavaScript SDK 已上線,原生支援 Claude Code、Codex、Gemini CLI。 ## 3. 技術差異化與競爭格局 競爭對手 | 核心模型 | 技術路線差異 ---|---|--- **DeepMind / Google** | AlphaFold 3 | 同為擴散架構;精度略高於 Boltz-2,但以 API 收費且禁商業用途(學術限定) **Chai Discovery** | Chai-1 | 同為開源擴散模型;應用廣度相近;無結合親和力預測 **Schrödinger** | FEP+ | 物理計算黃金標準;精度高但每目標需數天計算,Boltz-2 聲稱以 1000× 速度逼近其精度 **Recursion Pharmaceuticals** | 合作方(非直接競爭) | 提供 NVIDIA 算力協助訓練 Boltz-2;Recursion 自有 OS 模型與 Boltz 有互補 **Boltz 差異化定位** :以完全開源 + MIT 授權為稀缺門檻,同時快速商業化為全棧平台,聚焦「結構預測 + 結合親和力 + 生成設計」三位一體(競品多僅做其中一點);並以低廉 API 定價切入,對抗 Schrödinger 高授權費。 * * * ## 4. 應用場景與商業化路徑 **目標場景** : - 早期苗頭化合物(hit)發現(BoltzMol-1) - De novo 抗體 / 蛋白結合子設計(BoltzProt-1) - 小分子結合親和力虛擬篩選(Boltz-2) - 生物大分子複合物結構解析(Boltz-1/2) **商業化里程碑(公司已揭露或已發生)** : 1. 2024-11:Boltz-1 開源發布,GitHub 獲逾 4,100 顆星,建立科學社群 2. 2026-01:公司成立 + $28M 種子輪 + Pfizer 多年戰略合作公布 3. 2026-06-16:BoltzMol-1、BoltzProt-1、Boltz API 全棧發布 4. 2026-06-19:Takeda 部署 Boltz Lab 平台(本地端 + 雲端雙模式) * * * ## 5. 商業模式 Boltz 明確定位為**軟體與基礎設施提供商** ,非治療管線持有者,三大收入來源(公司揭露): 1. **Boltz API** :按次計費($0.025/預測起),適合高通量虛擬篩選 2. **Boltz Lab 訂閱** :多租戶雲端 / 單租戶雲端 / 全本地端三種部署選項,面向研究機構與藥廠 3. **企業定製模型** :以客戶私有資料精調基礎模型(Pfizer 合作模式)—— 公司保留基礎模型主權,客戶保留下游化合物歸屬 **已知客戶 / 合作夥伴** : - **Pfizer** (多年戰略合作,2026-01):定製結構預測、小分子親和力、生物製劑設計模型 - **Takeda** (平台部署,2026-06):全研究組織導入 Boltz Lab 本地端 + 雲端 - **Recursion Pharmaceuticals** :算力合作夥伴,協助訓練 Boltz-2 - **整合夥伴** :Benchling、Amazon Bio Discovery、Rowan、Phylo、Tamarind、Kiin Bio、Pauling.ai 等 商業條款(三項保證,公司揭露):用戶保留所有輸入 / 輸出的智財歸屬;客戶資料不用於 Boltz 重新訓練;資料隔離安全保障。 * * * ## 6. 科學文獻 Boltz-1: Democratizing Biomolecular Interaction Modeling — Wohlwend, Corso, Passaro et al., _bioRxiv_ 預印本, 2024-11(最新版 2025-05)。公司核心技術基礎,確立 Boltz-1 為首個以完全開源方式達到 AlphaFold3 水準精度的生物分子結構預測模型。 Boltz-2: Towards Accurate and Efficient Binding Affinity Prediction — Corso, Wohlwend, Passaro et al., _bioRxiv_ 預印本, 2025-06。Boltz-2 技術報告,描述聯合結構 + 結合親和力預測架構,首個深度學習模型逼近 FEP 精度且速度提升 1000 倍。 BoltzGen: Toward Universal Binder Design — Boltz 團隊, _bioRxiv_ 預印本, 2025-11。BoltzGen 生成式蛋白質設計模型,BoltzProt-1 前身,展示針對任意靶標設計結合子的通用框架。 Boltzina: Efficient and Accurate Virtual Screening via Docking-Guided Binding Prediction with Boltz-2 — [URL 查無頁面摘要,來自 arXiv 2508.17555], 2025-08。以 Boltz-2 為基礎的虛擬篩選工作流,展示 Boltz 技術在大規模藥物篩選中的衍生應用。 * * * ## 7. 專利發表 目前查無以 **Boltz PBC** 為主要權利人的已公告發明專利(早期新創、主力以學術預印本發布為主)。以下列替代指標佐證技術競爭力: * **開源影響力** :GitHub `jwohlwend/boltz` 儲存庫達 **4,100+ 顆星、850+ Fork** (MIT 授權,2025 年截至 v2.2.1),在生物分子預測開源工具中屬前列 * **頂尖機構採用** :Pfizer 與 Takeda 均在發布後數月內簽署正式合作,確認工業級可用性 * **平台整合廣度** :Benchling、Amazon Bio Discovery 等 8 家以上工具平台原生整合,反映生態滲透率 * **Regina Barzilay 實驗室背景** :指導教授為 MIT AI + 健康領域最具影響力研究者之一,確立學術公信力基礎 * * * ## 8. 近期財報彙整 [未上市,無公開財報] **已揭露私募輪次:** 輪次 | 金額 | 日期 | 主要投資人 ---|---|---|--- 種子輪 | $28M | 2026-01 | Andreessen Horowitz(a16z)、Amplify Partners、Zetta Venture Partners — | — | — | 天使:Clément Delangue(Hugging Face CEO)、Factorial Capital、Obvious Ventures **已公開營運指標:** - GitHub 儲存庫:4,100+ 顆星(模型社群採用指標)[查無公開資訊—確切 MAU/DAU] - 企業客戶:至少 Pfizer、Takeda 兩家藥廠(規模或 ARR [查無公開資訊]) - 平台整合:8 家以上第三方工具平台 - 團隊規模:[查無公開資訊](三位核心創辦人公開確認) ### 下一個關鍵執行里程碑 BoltzMol-1 命中化合物濕實驗室確效結果公布,以及 Pfizer/Takeda 合作模型的後續進展(預計 2026-H2);Boltz API 商業規模化(訂閱客戶數與 API 用量增長指標,公司至今未公開具體時間點)。[尚無明確時間點公開資訊]
www.biopetri.com
July 1, 2026 at 12:30 PM
PENALTY SHOOTOUT: D. Boltz (Chicago Fire II) — SCORED. Shootout: Chattanooga 2–1 Chicago Fire II. MLS Next Pro

#chattanooga #chicagofireii
June 28, 2026 at 1:06 AM
•ENRAGING?•

He DOES hav Exp!!… Real Estate Developer, exp!

His Daddy, Fred, put HIM in-charge of tearing down, WHOLE neighborhoods, in Levittown, NY-LI!

Dumald? Was so-far-REMOVED from Nutz&Boltz, Process? Rich/Lazy/GolfBoy, ‘LEARNED’ 1 thing:

How 2 BLAME •OTHERS•, when things went: Sideways!
May have? His car alone weighs 15000 pounds. It’s as if he’s had no experience building anything.
June 26, 2026 at 3:42 PM