Chai Discovery
The AI platform designing novel antibodies and biologics from scratch — compressing drug discovery from years to weeks.
Overview
Chai Discovery builds frontier AI models for de novo antibody and biologic design. Chai-1 (open-source, Apache 2.0) handles structure prediction; Chai-2 hits 16–20% experimental hit rates on novel targets — over 100x better than prior computational methods.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 11 Jul 2026 and may be out of date or incomplete.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.
- Commercial & academic use
- GitHub + PyPI package
- No-setup web server
- Proteins, DNA, RNA, ligands
- Zero-shot antibody design
- 16–20% experimental hit rate
- Bespoke model option
- Responsible Deployment vetting
Use cases
What you can produce with Chai Discovery
- Predicted 3D structures of protein-protein, protein-ligand, and antibody-antigen complexes (Chai-1)
- De novo CDR sequences for scFv, VHH nanobody, and minibinder antibody formats (Chai-2)
- Ranked antibody candidates with predicted binding affinity and drug-like physicochemical properties
- Validated binder hits at 16–20% experimental success rate across a 24-well-plate wet-lab workflow
- Custom bespoke model trained on client proprietary data (enterprise partnership tier)
- Open-source Python package (pip install chai-lab) for local GPU inference with full model weights
- Antibody-antigen complex structure predictions with DockQ scores exceeding 0.8 for 34% of test cases
- Roadmap/planned: bi-paratopic and bispecific/multi-specific antibody formats (described as future capability, not yet validated)
ASEAN Perspective
Chai Discovery in Southeast Asia
ASEAN pharmaceutical and biotech organisations — including institutions in Singapore's Biopolis, Thailand's National Center for Genetic Engineering and Biotechnology, and Indonesia's growing biotech sector — can access Chai-1 at no cost via the web interface or open-source GitHub release for academic and commercial structure prediction work. Chai-2 enterprise access, however, requires direct engagement with the company and carries pricing calibrated to large pharma budgets, creating a barrier for most ASEAN players without co-funding partnerships. Competition from ByteDance Research's Protenix (released publicly in February 2026 and reportedly exceeding AlphaFold 3 on several benchmarks) may present APAC teams with a regionally better-supported alternative for structure prediction. Chai's long-term ASEAN relevance will hinge on whether it opens lower-cost API tiers or forges academic consortium agreements in the region.
Chai Discovery is one of the most technically credible AI drug discovery platforms of 2025–2026. Chai-1's open-source Apache 2.0 release gives academic and commercial researchers immediate, zero-cost access to state-of-the-art biomolecular structure prediction rivalling AlphaFold 3 on key benchmarks, and the GitHub repository has attracted a growing research community. Chai-2's 16–20% experimental hit rate for de novo antibody design against 52 novel targets — validated across a 24-well-plate wet-lab workflow — represents a genuine step-change over prior computational methods that rarely exceeded 0.1%.
The major caveat is accessibility: Chai-2, the marquee product, is gated behind a "Responsible Deployment" framework with no public pricing, and enterprise access appears reserved for well-resourced biopharma partners willing to negotiate bespoke deals. The Eli Lilly partnership terms (mid-eight-figure annual fee) confirm the platform's value but also its cost ceiling for most organisations. For APAC biotech firms and academic groups, Chai-1 is an excellent free resource; Chai-2 enterprise access will depend on BD engagement with a San Francisco-headquartered team that has no confirmed regional office or APAC-specific programme to date.
What people say
Most AI drug-discovery startups talk about "compressing years into weeks." Chai Discovery has a wet-lab number to back it up: Chai-2 hit 16–20% experimental success designing antibodies against 52 targets that had zero known binders in the Protein Data Bank — more than 100 times better than the computational baselines that came before it.
Chai-1, the structure-prediction model, is the free half of the pitch: Apache 2.0 licensed, over 2,000 stars on GitHub, and cited in 300-plus papers within its first year — a legitimate AlphaFold 3 competitor that costs nothing to run. Chai-2 is the opposite story. It's gated behind a "Responsible Deployment" review, there's no public price, and the one number that leaked — Eli Lilly's mid-eight-figure annual fee from their January 2026 deal — tells you who this is actually built for.
A $130 million Series B in December 2025 pushed the company to unicorn status at a $1.3 billion valuation, with OpenAI among the backers. For a two-year-old company, that's a fast climb. For anyone outside big pharma or a well-funded biotech, Chai-1 is worth downloading; Chai-2 is a sales conversation, not a self-serve tool.
Summary of public user & expert reviews, compiled by RECATOOLS.
Notable facts
- Chai Discovery reached unicorn status ($1B+ valuation) in under two years from founding — faster than most biotech startups in history.
- CEO Joshua Meier co-developed ESM1 at Meta, one of the first transformer-based protein language models, before co-founding Chai at age ~30.
- Chai-1 is released under Apache 2.0, meaning a pharmaceutical company can legally run it commercially at zero royalty — something AlphaFold 3's licence explicitly restricts for commercial use.
- The Eli Lilly deal includes building a proprietary model trained on Lilly's internal data — a first-of-its-kind bespoke large-pharma AI collaboration disclosed at this scale.
Frequently asked questions
About this listing
This entry was compiled from publicly available data including Chai Discovery's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Chai Discovery 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 Chai Discovery directly →
Spotted something out of date? Suggest an update →
Chai Discovery in the news
AI & ML
ByteDance Launches Seedream 5.0 Pro, an Image Model That Outputs Editable Layers
Health & Wellness
German Quantum Health Startup NVision Raises US$55M to Combine Quantum MRI and Drug Discov...
AI & ML
EU AI Act Omnibus: Annex III High-Risk Deadline Extends 16 Months, Annex I Gets 12, Nudifi...
More in Research & Data
