Boltz
AI that designs the molecules of tomorrow for pharma and biotech companies.
Website: https://boltz.com/
Cover Block
Open sources
| Field | Value |
|---|---|
| Name | Boltz |
| Tagline | AI that designs the molecules of tomorrow for pharma and biotech companies. [boltz.com] |
| Headquarters | London, United Kingdom [PitchBook] |
| Founded | 2025 [Crunchbase, December 2025] |
| Stage | Seed [Crunchbase, December 2025] |
| Business Model | SaaS [boltz.com, retrieved 2026] |
| Industry | Deeptech |
| Technology | AI / Machine Learning |
| Geography | Global / Remote-First |
| Growth Profile | Venture Scale |
| Funding Label | $28M Seed Round [PR Newswire, June 2026] |
| Total Disclosed | ~$28,000,000 [PR Newswire, June 2026] |
Links
Open sources
- Website: https://boltz.com/
- GitHub: https://github.com/jwohlwend/boltz
What an Investor Needs First
PUBLIC Boltz builds AI software for molecular design, aimed at pharma and biotech teams that want to generate and optimize small molecules and biologics with fewer experimental cycles, and it merits attention now because it launched publicly with a disclosed $28 million seed round and a full-stack product push in June 2026 [boltz.com, retrieved 2026] [Boltz PBC Launches with $28M to Democratize AI Platforms for Drug Discovery, 2026] [Announcing BoltzMol-1, BoltzProt-1, and the Boltz API, retrieved 2026]. The public record is still thin on the founding story and team composition, but Boltz presents itself as a London-headquartered public benefit corporation founded in 2025, and founder-linked public profiles captured in the source set point to Gabriele Corso, Jeremy Wohlwend, and Saro Passaro as co-founders, with Corso identified publicly as CEO on LinkedIn [Boltz PBC Launches with $28M to Democratize AI Platforms for Drug Discovery, 2026] [linkedin.com, retrieved 2026].
The product stack is broader than a single model: Boltz Lab is the end-to-end platform, the Boltz API is the integration layer, and the company says customers can access proprietary models such as BoltzProt-1 and BoltzMol-1 alongside open-source offerings including Boltz-2 and BoltzGen, with deployment options spanning multi-tenant cloud, dedicated environments, and on-premises setups [boltz.com, retrieved 2026] [boltz.com, retrieved 2026]. What appears differentiated, based on public materials, is the attempt to pair open scientific model adoption with enterprise packaging, security positioning, and workflow delivery for non-specialist R&D users, rather than selling only research artifacts or only infrastructure [Science at the speed of inference - by Gabriele Corso, retrieved 2026] [Boltz PBC Launches With $28M Seed Round, Pfizer Collaboration, and Three Guarantees, retrieved 2026] [boltz.com, retrieved 2026].
Commercially, the company fits a venture-scale SaaS profile, though most operating metrics remain company-supplied and should be treated cautiously; public materials claim broad usage among scientists and companies, while third-party databases are firmer only on the seed financing and a reported 37-employee footprint [boltz.com, retrieved 2026] [PitchBook, retrieved 2026]. The cap table appears strong for a first institutional round, with Zetta Ventures, Amplify Partners, a16z, Obvious Ventures, and Factorial Capital listed as backers across the source set, but the precise lead investor is not consistently confirmed in the materials provided [Boltz Funding, Valuation & Investors, retrieved 2026] [Tracxn, retrieved 2026] [Boltz PBC Launches with $28M to Democratize AI Platforms for Drug Discovery, 2026].
Over the next 12 to 18 months, the central watchpoints are whether Boltz can convert scientific credibility into durable enterprise adoption, whether collaborations such as those announced with Pfizer and Takeda translate into repeatable platform usage, and whether the company can validate product claims with more independent evidence than company channels currently provide [prnewswire.com, October 2026] [prnewswire.com, retrieved 2026]. The setup is promising, particularly in a category where model quality, distribution, and deployment trust all matter, but the present public evidence base remains uneven and still leans heavily on company disclosures.
Partially corroborated -- Core company description, product launches, and funding are corroborated by company materials and external databases, but several material claims, especially traction and product performance, remain company-led.
Taxonomy Snapshot
| Axis | Value |
|---|---|
| Stage | Seed |
| Business Model | SaaS |
| Industry / Vertical | Deeptech |
| Technology Type | AI / Machine Learning |
| Geography | Global / Remote-First |
| Growth Profile | Venture Scale |
| Funding | $28M Seed Round |
Inside the Company
PUBLIC
Boltz presents as a newly formed biomolecular AI company with an unusually public product posture for its age. The company describes itself as building "AI that designs the molecules of tomorrow" for pharmaceutical and biotech users, and its website positions the business around molecular design software rather than a services-led discovery model [boltz.com]. Structured company data places headquarters in London and a 2025 founding year, while external startup databases cited in the research record also list the company as venture-backed at seed stage [Crunchbase, December 2025].
The legal form is clearer than the founding narrative. Company materials identify Boltz as a public benefit corporation, or PBC, and describe it as an applied AI research company focused on biomolecular models and software infrastructure for scientific discovery [PR Newswire, 2026] [boltz.com]. That framing matters because the business appears to be trying to commercialize an open-model research base into enterprise software and platform access, a structure that is more common in frontier AI than in traditional biotech tooling [boltz.com].
The short public timeline is straightforward. In 2025, the company appears in startup databases as founded and seed-stage [Crunchbase, December 2025]. On June 16, 2026, Boltz publicly launched BoltzMol-1, BoltzProt-1, and the Boltz API, alongside the introduction of Boltz Lab as its end-to-end molecular design platform [boltz.com] [Labcritics, June 2026]. Later in 2026, the company announced strategic collaborations with Pfizer and Takeda, which suggests an early push to validate the platform inside large pharmaceutical research organizations rather than only through self-serve adoption [PR Newswire, October 2026] [PR Newswire].
Partially corroborated -- Confirmed in part by the company website and Crunchbase; legal entity and milestone details rely materially on company and company-distributed announcements.
Under the Hood
MIXED Boltz is positioning itself as a full-stack biomolecular design software company, but the public record is still much stronger on product surface area than on independently verified performance. On its website, the company describes Boltz Lab as a platform for end-to-end molecular design, with both an interface for scientists and an API for integration into external workflows [boltz.com]. The named product set includes BoltzProt-1 for protein design, BoltzMol-1 for small-molecule work, and the open-source models Boltz-2 and BoltzGen [boltz.com]. A June 16, 2026 launch date for BoltzMol-1, BoltzProt-1, and the Boltz API is corroborated across the company announcement and third-party writeups, which is one of the cleaner product facts in the file [Announcing BoltzMol-1, BoltzProt-1, and the Boltz API] [Labcritics, 2026] [Latent Space].
The technical pitch rests on a mix of open models, proprietary commercial layers, and enterprise deployment controls. Public materials say the platform supports multi-tenant cloud, single-tenant dedicated environments, and on-premises deployment, and that customer data is kept separate from model training [boltz.com]. Boltz also claims SOC 2 Type 1 verification, inference times of about 15 seconds, and infrastructure with thousands of GPUs on demand, although those points remain company-sourced in this dataset [boltz.com]. On the model side, the open-source Boltz-1 work is framed as approaching AlphaFold3-level accuracy in biomolecular structure prediction, while BoltzMol-1 and BoltzProt-1 are presented as commercially packaged systems for hit discovery and protein design rather than research artifacts alone [GitHub] [Boltz-1: The First Open-Source Model to Approach AlphaFold3 Accuracy] [boltz.com].
What is notable here is less the existence of another AI drug discovery interface and more the attempt to bridge open scientific credibility with enterprise usability. The Excelra partnership adds one concrete signal on data inputs, with GOSTAR data cited as training material for ADMET models inside BoltzMol-1 [PR Newswire, 2026]. Separately, Boltz says the API is accessible through partners including Benchling, Phylo Bio, Tamarind Bio, Pauling AI, Mirror Physics, and Cultivarium FRX, which, if sustained, would matter because workflow embedment is usually a better indicator of durability than raw model downloads [SynBioBeta on X, 2026]. Still, most efficacy claims, including hit rates and usage breadth, are company-authored or promotional, so the current public evidence supports product breadth more clearly than validated technical advantage [boltz.com] [Announcing BoltzMol-1, BoltzProt-1, and the Boltz API].
Partially corroborated -- Product names, launch timing, and core platform structure are corroborated by company materials and third-party coverage, but most performance and adoption claims in this section remain company-sourced.
Market Research
PUBLIC
This market matters now because Boltz sits at the intersection of two spending pools that are already real, drug discovery software and AI infrastructure for life sciences, even if the company's own addressable market has not been sized in the public materials provided [boltz.com, retrieved 2026].
The cleanest way to frame market size here is by analogy rather than by pretending the evidence is tighter than it is. Boltz sells AI models and workflow software for molecular design to pharma and biotech teams [boltz.com, retrieved 2026]. The source set for this report does not include a named third-party TAM for AI-native molecular design software, so the market must be triangulated through adjacent categories: drug discovery platforms, computational biology software, and enterprise AI tooling sold into R&D organizations. That matters because Boltz is not marketing a point model alone. It presents a stack that includes open-source models, proprietary models, an API, and a platform called Boltz Lab, which suggests a budget line closer to research software and scientific compute than to a single-purpose informatics tool [boltz.com, retrieved 2026].
Public demand signals in the materials are directionally strong, but most are still company-supplied and should be read that way. Boltz claims its models have 5M+ downloads, 100k+ scientists, 1k+ companies, and usage at top 20 pharma companies [boltz.com, retrieved 2026]. Separately, the company announced collaborations with Pfizer and Takeda in 2026, which at minimum indicates that large pharmaceutical organizations are willing to evaluate or deploy biomolecular AI models through formal partnerships rather than only through academic experimentation [PR Newswire, October 2026] [PR Newswire]. A third demand driver is usability: Boltz's public positioning emphasizes faster inference, API access, and deployment choices spanning multi-tenant cloud, dedicated environments, and on-premises setups, all of which map to practical enterprise adoption hurdles in regulated R&D settings [boltz.com, retrieved 2026].
The adjacent markets are broad enough that substitution risk cuts both ways. On one side, Boltz can benefit from budgets already allocated to structure prediction, molecular modeling, screening, wet-lab prioritization, and scientific cloud compute [boltz.com, retrieved 2026]. On the other, buyers may treat parts of the product as substitutable with internal computational chemistry teams, open-source biomolecular models, or incumbent research software paired with general-purpose AI infrastructure. The open-source heritage of Boltz-1 and the presence of newer proprietary surfaces such as BoltzMol-1 and BoltzProt-1 make that boundary especially important: the commercial opportunity likely rests less on raw model access than on workflow integration, enterprise deployment, experimental validation claims, and data advantage such as the Excelra GOSTAR relationship cited by the company and its partner [GitHub] [PR Newswire].
Macro and regulatory forces are also unusually relevant here because the customer base is conservative by design. Boltz states that customer data remains confidential, is not used to train models, and can be deployed on-premises or in single-tenant environments, while also stating SOC 2 Type 1 verification on its site [boltz.com, retrieved 2026]. Those are product details, but they are also market signals: life sciences buyers increasingly need procurement-ready AI systems that fit existing IP controls, validation processes, and security review. If that requirement hardens, it should favor vendors that can package frontier models into controlled enterprise deployments. If it softens, more value may accrue to open-source distribution and in-house model operations instead.
| Market lens | Public evidence | Relevance to Boltz |
|---|---|---|
| Drug discovery AI workflows | Boltz markets end-to-end molecular design through Boltz Lab and API [boltz.com, retrieved 2026] | Suggests platform-level budget capture rather than single-model monetization |
| Enterprise pharma adoption | Collaborations announced with Pfizer and Takeda in 2026 [PR Newswire, October 2026] [PR Newswire] | Indicates buyer interest from large regulated R&D organizations |
| Open-source biomolecular models | Boltz-1 described as approaching AlphaFold3 accuracy in open source materials [GitHub] [CBIRT] | Expands top-of-funnel adoption, but also raises questions about where proprietary monetization begins |
| Scientific compute and secure deployment | Boltz advertises multi-tenant, single-tenant, and on-premises deployment, plus SOC 2 Type 1 [boltz.com, retrieved 2026] | Aligns with security and infrastructure requirements common in pharma procurement |
The market evidence here points to a category with real buyer urgency, but not yet to a well-bounded public TAM for Boltz specifically. The stronger read is qualitative: enterprise life sciences appears willing to spend on AI systems that shorten experimental cycles, provided they arrive with security, deployment flexibility, and workflow fit.
Claim stands unchecked -- This section relies primarily on company materials from boltz.com and company-distributed partnership announcements, with no third-party market sizing report in the provided source set.
Competition and Substitutes
Competitive map
MIXED Boltz is positioning itself less as a single-model research project and more as a full-stack molecular design platform, which puts it in competition with both biomolecular model builders and the software layers that sit between those models and pharma workflows [boltz.com, retrieved 2026] [PR Newswire, October 2026].
The public record here is thinner than the company narrative suggests, so the competitive map has to start with what is actually named. On the model layer, Boltz repeatedly benchmarks itself against AlphaFold3 through claims that Boltz-1 was the first open-source model to approach that accuracy level for biomolecular structure prediction [GitHub, retrieved 2026] [CBIRT, retrieved 2026] [boltz.com, retrieved 2026]. Chai-1 appears in those benchmark materials as a direct technical comparison point, with Boltz-1 reported at 65 percent LDDT-PLI on CASP15 versus 40 percent for Chai-1, and 83 percent on DockQ>0.23 versus 76 percent for Chai-1, though those figures come from model materials rather than an independent benchmark publisher [boltz.com, retrieved 2026].
That implies three practical competitor groups. First are frontier model incumbents, including AlphaFold3 by reference and Chai-1 by explicit benchmark comparison, where the contest is scientific performance, openness, and commercial usability rather than classic enterprise distribution [boltz.com, retrieved 2026] [GitHub, retrieved 2026]. Second are platform challengers and workflow providers, where Boltz Lab and the Boltz API compete to become the operating layer through which medicinal chemistry and protein design teams actually run programs, integrate tools, and move candidates through discovery [boltz.com, retrieved 2026]. Third are adjacent substitutes inside large pharma, namely internal computational biology teams and stitched-together vendor stacks, which may be slower but already sit inside validated procurement, security, and data-governance processes [PR Newswire, retrieved 2026].
Boltz's edge today appears to rest on a combination of open-source adoption, commercial packaging, and early enterprise signaling, but most of that edge remains at least partly company-described. The strongest public evidence is that the company has turned its research assets into a broader product surface: open-source models including Boltz-2 and BoltzGen, proprietary models including BoltzMol-1 and BoltzProt-1, a user-facing platform in Boltz Lab, and an API that is already described as accessible through partners such as Benchling, Phylo Bio, Tamarind Bio, Pauling AI, Mirror Physics, and Cultivarium FRX [boltz.com, retrieved 2026] [SynBioBeta, June 2026]. The collaboration announcements with Pfizer and Takeda matter here because they suggest Boltz is not competing only for mindshare among academic users; it is also trying to secure credibility with major pharmaceutical research organizations, which is where platform stickiness could become durable if deployments deepen beyond pilot scope [PR Newswire, October 2026] [PR Newswire, retrieved 2026].
That said, durability is not the same thing as current proof. Open-source adoption can seed distribution, and Boltz cites more than 100,000 scientists using its models or related tools, with millions of downloads claimed on its site, but those figures are largely company-sourced and mix research adoption with enterprise product usage [boltz.com, retrieved 2026] [Substack, retrieved 2026]. The more defensible near-term edge may be narrower: Boltz is trying to package high-performance biomolecular models into a deployable enterprise product with cloud, single-tenant, and on-premises options, plus a stated SOC 2 Type 1 posture and customer IP protections [boltz.com, retrieved 2026]. That edge is durable only if the company can show that packaging, security, and experimental hit rates translate into repeat procurement, because raw model differentiation in AI biology tends to erode as papers, weights, and techniques diffuse.
The main exposure is that named alternatives can still win on either scientific prestige or workflow control. AlphaFold3 remains the reference standard Boltz chooses to compare itself against, which means Boltz benefits from that framing only so long as performance remains close enough and commercial access remains more attractive on Boltz's side [boltz.com, retrieved 2026]. Chai-1, while only visible here through Boltz's own benchmark framing, has the obvious advantage of already being the comparison target in the same technical conversation, and tools like Benchling sit closer to the system of record for many life-science teams even if they are not foundational-model vendors themselves [boltz.com, retrieved 2026] [SynBioBeta, June 2026]. Boltz is also exposed in channels it does not fully own: if partner integrations become the main path to usage, those partners can capture workflow gravity while Boltz supplies intelligence underneath.
The most plausible 18-month scenario is a split market rather than a single winner. Boltz is the likely winner if enterprise buyers decide that open-model credibility plus secure deployment plus an end-to-end product layer is more valuable than using standalone research models or internally assembled tooling, particularly if Pfizer- and Takeda-style collaborations convert into broader organizational rollout [PR Newswire, October 2026] [PR Newswire, retrieved 2026]. Chai-1 is the plausible loser if benchmark-based differentiation narrows and customers prioritize deployment, integrations, and validated discovery workflows over a model that is evaluated primarily at the technical layer [boltz.com, retrieved 2026]. The reverse is also true, and it is the key competitive caution: if buyers keep treating this category as a model bake-off rather than a workflow platform decision, the advantage shifts back toward whichever research system is seen as closest to the scientific frontier.
Opportunity
Upside case
PUBLIC The prize here is unusually large if Boltz can turn early model adoption into durable enterprise workflows, because drug discovery software that shortens experimental cycles can move from a research tool budget to a strategic platform budget inside large pharma and biotech organizations [boltz.com, retrieved 2026] [PR Newswire, October 2026] [PR Newswire].
The headline opportunity is to become a default biomolecular design layer used across both internal research teams and external software ecosystems. That outcome is not yet proven, but it is at least reachable on the public record. Boltz has paired an open-source entry point, including Boltz-2 and BoltzGen, with proprietary models such as BoltzProt-1 and BoltzMol-1, then wrapped those models in Boltz Lab and an API that can be embedded into existing scientific workflows [boltz.com, retrieved 2026] [Announcing BoltzMol-1, BoltzProt-1, and the Boltz API] [Science at the speed of inference - by Gabriele Corso]. The company also has public evidence of enterprise-facing distribution through collaborations with Pfizer and Takeda, plus API availability through partners including Benchling and other computational biology tools, which matters because platform winners in technical software usually arrive through workflow integration rather than direct seat sales alone [PR Newswire, October 2026] [PR Newswire] [SynBioBeta on X, 2026].
A second reason the upside is real rather than purely conceptual is that Boltz appears to be pursuing both adoption breadth and product depth at the same time. Its public materials describe more than 100,000 scientists using the open-source models, while the commercial stack adds secure deployment options, on-premises support, and claims around customer IP ownership and data isolation that address the procurement concerns of pharma buyers [Science at the speed of inference - by Gabriele Corso] [boltz.com, retrieved 2026]. That combination, open model reach on one side and enterprise controls on the other, is a credible route to becoming infrastructure rather than a point solution.
| Scenario | What happens | Catalyst | Why it's plausible |
|---|---|---|---|
| Open-to-enterprise conversion | Boltz turns broad open-source model adoption into paid usage of Boltz Lab, BoltzProt-1, BoltzMol-1, and the API across research organizations | The June 16, 2026 launch of BoltzMol-1, BoltzProt-1, and the API gives users a direct upgrade path from research model usage to commercial workflows [Announcing BoltzMol-1, BoltzProt-1, and the Boltz API] | Public materials claim 100,000-plus scientists use the model family, and the company now offers enterprise deployment modes and workflow software rather than only research artifacts [Science at the speed of inference - by Gabriele Corso] [boltz.com, retrieved 2026] |
| Embedded infrastructure layer | Boltz becomes a recurring model and inference provider inside the core software stack used by pharma and biotech teams | API distribution through partners including Benchling broadens reach beyond direct sales, while strategic collaborations with Pfizer and Takeda validate enterprise relevance [SynBioBeta on X, 2026] [PR Newswire, October 2026] [PR Newswire] | Scientific software markets often concentrate around tools that sit inside existing systems of record; Boltz has public evidence of pursuing that exact insertion point through APIs and partners [boltz.com, retrieved 2026] [SynBioBeta on X, 2026] |
| Full-stack discovery platform | Boltz advances from model vendor to end-to-end discovery platform spanning small molecules, proteins, screening, and optimization | The launch of Boltz Lab with small-molecule and protein design agents, plus the Excelra GOSTAR data partnership, expands from prediction into workflow execution and property filtering [Science at the speed of inference - by Gabriele Corso] [PR Newswire] | The company already markets a platform for end-to-end molecular design and cites integrated ADMET triage via GOSTAR-backed training data, which is the sort of stack expansion that can raise switching costs over time [boltz.com, retrieved 2026] [PR Newswire] |
What compounding looks like is straightforward, even if the public evidence is still early. Open-source adoption creates awareness among scientists, that awareness lowers the friction of getting a team to trial Boltz Lab or the API, paid deployments generate proprietary workflow data and customer feedback, and those deployments can make the next enterprise sale easier if the models keep improving in practical tasks such as hit discovery and developability filtering [Science at the speed of inference - by Gabriele Corso] [Announcing BoltzMol-1, BoltzProt-1, and the Boltz API] [boltz.com, retrieved 2026]. The partner layer matters as well. If tools such as Benchling and other listed partners expose the API inside daily research environments, distribution can compound without Boltz owning every end-user relationship directly [SynBioBeta on X, 2026].
There is also an early data and workflow moat argument, although it should be stated carefully because the public record does not show proprietary outcome datasets at scale. Boltz has announced an enterprise data partnership with Excelra's GOSTAR for ADMET model training, and its website emphasizes secure deployment and customer data separation, which could make high-value customers more willing to run sensitive programs through the system [PR Newswire] [boltz.com, retrieved 2026]. If that trust converts into long-lived deployment inside major drug developers, the company could accumulate product knowledge, integration depth, and validation loops that are hard for a pure open-source project to match.
The size of the win is best framed as a scenario, not a forecast. No credible public market or TAM source was provided in the research set, so the cleanest benchmark is the strategic importance implied by enterprise collaborations with Pfizer and Takeda and the breadth implied by 100,000-plus scientists using the model family [PR Newswire, October 2026] [PR Newswire] [Science at the speed of inference - by Gabriele Corso]. If Boltz becomes the embedded model and workflow layer for a meaningful share of large biopharma discovery teams, the outcome could support a multibillion-dollar software and infrastructure business (scenario, not a forecast). That upside case depends less on being the best standalone model on any single benchmark, and more on becoming the system through which molecular design work is actually done.
Claim stands unchecked -- This section relies materially on company materials and company-linked announcements, with limited independent public corroboration of adoption, performance, and commercial traction.
Sources
Open sources
[boltz.com] Boltz | https://boltz.com/
[boltz.com, retrieved 2026] Boltz | https://boltz.com/
[Crunchbase, December 2025] Boltz Funding, Valuation & Investors | https://www.crunchbase.com/organization/boltz
[PR Newswire, June 2026] Boltz PBC Launches with $28M to Democratize AI Platforms for Drug Discovery | https://www.prnewswire.com/news-releases/boltz-pbc-launches-with-28m-to-democratize-ai-platforms-for-drug-discovery-302482221.html
[Announcing BoltzMol-1, BoltzProt-1, and the Boltz API, retrieved 2026] Announcing BoltzMol-1, BoltzProt-1, and the Boltz API | https://boltz.com/blog/boltzmol-1-boltzprot-1-and-the-boltz-api
[linkedin.com, retrieved 2026] Gabriele Corso - Co-founder and CEO at Boltz | PhD at MIT | https://uk.linkedin.com/in/gcorso
[Science at the speed of inference - by Gabriele Corso, retrieved 2026] Science at the speed of inference | https://gabrielecorso.substack.com/p/science-at-the-speed-of-inference
[Boltz PBC Launches With $28M Seed Round, Pfizer Collaboration, and Three Guarantees, retrieved 2026] Boltz PBC Launches With $28M Seed Round, Pfizer Collaboration, and Three Guarantees | https://www.latent.space/p/boltz
[PitchBook, retrieved 2026] Boltz 2026 Company Profile: Valuation, Funding & Investors | https://pitchbook.com/profiles/company/boltz
[Boltz Funding, Valuation & Investors, retrieved 2026] Boltz Funding, Valuation & Investors | https://www.crunchbase.com/organization/boltz
[Tracxn, retrieved 2026] Boltz - 2026 Funding Rounds & List of Investors | https://tracxn.com/d/companies/boltz
[prnewswire.com, October 2026] Boltz and Pfizer Announce Strategic Collaboration to Develop and Deploy State-of-the-Art Biomolecular AI Foundation Models | https://www.prnewswire.com/news-releases/boltz-and-pfizer-announce-strategic-collaboration-to-develop-and-deploy-state-of-the-art-biomolecular-ai-foundation-models-302656405.html
[PR Newswire] Boltz Announces Collaboration with Takeda to Deploy Frontier Biomolecular AI Models Across Takeda's Research Organization | https://www.prnewswire.com/news-releases/boltz-announces-collaboration-with-takeda-to-deploy-frontier-biomolecular-ai-models-across-takedas-research-organization-302804082.html
[PR Newswire, 2026] Boltz Announces Collaboration with Takeda to Deploy Frontier Biomolecular AI Models Across Takeda's Research Organization | https://www.prnewswire.com/news-releases/boltz-announces-collaboration-with-takeda-to-deploy-frontier-biomolecular-ai-models-across-takedas-research-organization-302804082.html
[Labcritics, June 2026] BoltzMol-1, BoltzProt-1 and the Boltz API: AI Drug Discovery Goes Full-Stack | https://www.labcritics.com/post/boltzmol-1-boltzprot-1-and-the-boltz-api-ai-drug-discovery-goes-full-stack
[GitHub] Official repository for the Boltz biomolecular interaction models | https://github.com/jwohlwend/boltz
[Boltz-1: The First Open-Source Model to Approach AlphaFold3 Accuracy] Boltz-1: The First Open-Source Model to Approach AlphaFold3 Accuracy | https://boltz.bio/blog/boltz-1
[CBIRT] MIT Unveils Boltz-1: An Open-Source Tool for Biomolecular Structure Prediction | https://cbirt.net/mit-unveils-boltz-1-an-open-source-tool-for-biomolecular-structure-prediction/
[SynBioBeta on X, 2026] @boltz_bio has launched BoltzMol-1, BoltzProt-1, and the Boltz API | https://x.com/SynBioBeta/status/1934983387082508476
[Substack, retrieved 2026] Science at the speed of inference | https://gabrielecorso.substack.com/p/science-at-the-speed-of-inference
Articles about Boltz
- Boltz's Open-Source Models Anchor a $28 Million Bet on AI Drug Discovery — The public-benefit corporation's new Boltz Lab platform packages frontier models for protein and small-molecule design, backed by a seed round from a16z and others.