Boltz

First fully open-source model to approach AlphaFold3-level structure prediction, MIT-licensed

What Boltz does

  • First fully open-source model to approach AlphaFold3-level structure prediction, MIT-licensed
  • Boltz-2 adds binding-affinity prediction, not just structure
  • Hostable free via community web apps (e.g. Tamarind Bio) or run locally

Boltz — straight answers

What is Boltz?

Boltz is listed under Science & Research AI, in the Research & Learning category on Flocci AI Tools. First fully open-source model to approach AlphaFold3-level structure prediction, MIT-licensed. It is free, with no paid plan attached, and it lives at github.com.

Is Boltz free?

Boltz is listed as fully free — there is no paid tier attached to it in the catalog. That makes it one of the 396 entries on Flocci AI Tools with no upgrade path built in.

Can I use Boltz in a browser, or is there an app?

Boltz needs to be installed or self-hosted — it doesn't run in a browser tab. It is available on self-hosting and the command line.

What can Boltz do?

Boltz does 3 things the catalog singles out: First fully open-source model to approach alphafold3-level structure prediction, mit-licensed; boltz-2 adds binding-affinity prediction, not just structure; hostable free via community web apps (e.g. tamarind bio) or run locally.

What is the best free alternative to Boltz?

AlphaFold Server is the closest free alternative: it sits in the same Science & Research AI sub-category and is free. Biomni, AI Scientist-v2 and Evo 2 also start free. The full list is on the alternatives page.

See the full list →

AlphaFold Server

Science & Research AI
free
  • Free browser-based access to AlphaFold 3-class structure prediction for non-commercial research
  • Predicts structure/interactions of proteins, DNA, RNA, ligands in one model
  • Backed by the AlphaFold Protein Structure Database of 200M+ predicted structures

Chai-1 (Chai Discovery)

Science & Research AI
freemium
  • Chai-1 is free, Apache-2.0-licensed for academic and commercial use — a genuinely open structure-prediction model
  • Chai-2 (paid, enterprise) generates de novo antibodies directly from target epitopes, used by Lilly, Novartis, Pfizer
  • Autonomous AI agent that runs literature review + data analysis cycles, condensing ~6 months of research into a day per vendor
  • Publicly available via web interface and API, not an internal-only research demo
  • Kosmos itself requires a paid subscription; the free plan covers only standard agents.

Biomni

Science & Research AI
freeNew
  • General-purpose biomedical agent combining LLM planning with code execution
  • Integrates 150+ bioinformatics tools and 59 databases
  • Web interface so biologists can use it without coding
  • Open-source from Stanford SNAP lab

Evo 2

Science & Research AI
freeNew
  • DNA language model at single-nucleotide resolution with 1M base-pair context
  • Variant effect prediction, sequence completion and genome design
  • Free hosted 40B access via NVIDIA build.nvidia.com
  • Apache 2.0 code, weights and training data open; published in Nature

AI Scientist-v2

Science & Research AI
freeNew
  • Agentic tree search proposes ideas, runs experiments and writes papers
  • No human-authored templates needed
  • Open code from Sakana AI
  • Requires disclosure of AI use in resulting papers

Head-to-head comparisons