Isomorphic Labs IsoDDE
Isomorphic Labs' in-house drug design engine, not available to buy
Overview
IsoDDE (Isomorphic Labs Drug Design Engine) is the unified AI system Alphabet's Isomorphic Labs uses to design drug candidates, covering protein-ligand and antibody structure prediction, binding affinity and pocket finding. It was unveiled in February 2026 with a vendor technical report. It is not sold or licensed; outside access is only through research collaborations such as those with Eli Lilly, Novartis and Johnson & Johnson.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 14 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- Multi-target drug-design collaborations
- Named partners: Eli Lilly, Novartis, Johnson & Johnson
- Partnering contact listed on the vendor site
Use cases
What you can produce with Isomorphic Labs IsoDDE
- A published technical report benchmarking IsoDDE against AlphaFold 3, Boltz-2 and FEP+ (CC BY 4.0)
- Protein-ligand structure predictions for novel pockets and ligands, for Isomorphic and its partners
- Antibody-antigen interface and CDR-H3 loop models
- Binding-affinity rankings used to prioritise molecules in design programmes
- Candidate binding pockets, including cryptic ones, predicted from sequence alone
- Drug candidates designed for collaborators across small molecules and biologics
ASEAN Perspective
Isomorphic Labs IsoDDE in Southeast Asia
ASEAN's pharmaceutical and biotech sectors — including Singapore's Biopolis cluster, Malaysia's NIH, and Indonesia's emerging drug innovation programs — have no direct access pathway to IsoDDE, which is reserved for top-tier global pharma partners. Notably, Temasek (Singapore's sovereign wealth fund) participated in the May 2026 Series B round, signalling regional institutional confidence in Isomorphic Labs' long-term value even without local access. The indirect benefit for ASEAN may come through any licensed drugs eventually commercialised by Novartis, Eli Lilly, or J&J using IsoDDE-designed compounds. ASEAN researchers seeking comparable open tools should look to AlphaFold 3 (free for non-commercial use) or Boltz-2 (MIT licence) as accessible alternatives.
IsoDDE is worth knowing about because of who built it and what Isomorphic says it can do: a self-reported doubling of AlphaFold 3's success on the hardest protein-ligand cases, stronger antibody modelling, and affinity estimates competitive with physics-based methods. The technical report is open to read, but the results have not been reproduced outside the company.
It is not something to evaluate as a tool. There is no API, licence or academic access, and the only route in is a negotiated pharmaceutical partnership. Read the report; for work you can actually run, use openly available structure-prediction models.
What we found
IsoDDE, the Isomorphic Labs Drug Design Engine, is not a product anyone can buy. Isomorphic Labs, the London drug-discovery company Demis Hassabis set up as an Alphabet subsidiary, unveiled it on 10 February 2026 as the unified set of models its own scientists use to design medicines. It is also applied in research collaborations with Eli Lilly, Novartis and Johnson & Johnson. There is no API, licence, free tier or academic access, and no code or model weights were released.
The headline claim, that IsoDDE more than doubles the accuracy of AlphaFold 3, comes from Isomorphic's own 27-page technical report on Zenodo, not from an outside evaluation. It is a specific claim. On the hardest slice of the public Runs N' Poses benchmark, the protein-ligand pairs least like anything in training, IsoDDE succeeded on 50.0% of 60 cases against 23.3% for AlphaFold 3, and the report's own bootstrapped interval puts the gain anywhere from about 8 to 43 points. Filtering and clustering that set narrows the result to 37.2% against 16.7% across 39 cases. The report also claims a 2.3-fold gain over AlphaFold 3 on high-accuracy antibody-antigen interfaces, binding-affinity correlations above the physics-based FEP+ method on one public set, and pocket finding about 1.5 times better than the open-source P2Rank.
The money behind it is large. Isomorphic raised $600 million in March 2025, led by Thrive Capital, then $2.1 billion in a Series B announced on 12 May 2026, again led by Thrive, with Alphabet, GV, MGX, Temasek, CapitalG and the UK Sovereign AI Fund. The January 2024 Lilly and Novartis deals carried $82.5 million upfront between them and close to $3 billion in potential milestones, by press accounts; terms for Johnson & Johnson, signed in January 2026, were not disclosed.
No Isomorphic drug had entered a clinical trial by mid-October 2026; a September post said its preclinical data was helping it gear up for clinical development. For anyone outside those partnerships, IsoDDE is a report to read, not a tool to run.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
Notable facts
- Demis Hassabis won the 2024 Nobel Prize in Chemistry for AlphaFold 2 — the same foundational work that underpins IsoDDE.
- IsoDDE achieves 39% high-fidelity antibody-antigen prediction accuracy versus only 2% for the open-source Boltz-2 model.
- Singapore's Temasek joined the $2.1B Series B — making it one of the few ASEAN institutions with a financial stake in the platform.
- Isomorphic Labs' pharma partnerships with Novartis and Eli Lilly alone carry potential milestone payments of up to ~$3 billion combined.
Frequently asked questions
About this listing
This entry was compiled from publicly available data including Isomorphic Labs IsoDDE's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Isomorphic Labs IsoDDE unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Isomorphic Labs IsoDDE directly →
Spotted something out of date? Suggest an update →
More in Research & Data
