TxGemma
Open Gemma-based models for therapeutic development
About
TxGemma is a suite of open models from Google DeepMind, announced on March 25, 2025, fine-tuned from Gemma 2 in 2B, 9B and 27B sizes for therapeutics. Trained on over 7 million therapeutic examples from the Therapeutics Data Commons, it covers target identification, property prediction and clinical-trial reasoning, with chat variants that explain their predictions. DeepMind reports the 27B model beats or matches the prior state of the art on 64 of 66 tasks. It ships under the Gemma terms rather than a fully open license, and predictions are research aids, not clinical guidance.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Team-reported: TxGemma-27B Predict beats or matches prior best on 64 of 66 therapeutic tasks, outperforming specialist models on 26.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Google DeepMind release with official model card and preprint; Benchmarks are DeepMind-reported on 66 TDC tasks; Predictions are research aids, not clinical guidance. License: Gemma Terms of Use (custom, per Google). Not a fully open license..
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 8.0/10
- Openness · license + docs
- 7.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · Google DeepMind release with official model card and preprint
- · Benchmarks are DeepMind-reported on 66 TDC tasks
- · Predictions are research aids, not clinical guidance