Living Models / Google DeepMind · AI For Plant Genomics

Living Models: Gemma 4 and BOTANIC-1

A Google Gemmaverse case study published 6 October. Living Models, an AI-biology lab in Paris, pairs Gemma 4 E4B with BOTANIC-1, its foundation model pretrained on DNA from 320 plant species, to find the mutations behind crop traits. Gemma 4 acts as the agent: it writes code, runs bcftools and Ensembl VEP, and sends plain-language questions to BOTANIC-1. In a melon test the system ranked the known causal mutation first out of 2,494 candidates in under four minutes, and across 20 runs it found it first 90% of the time, against 15% for expert-guided classical analysis, vendor-reported. The workflow runs locally on one NVIDIA L4 GPU through Ollama, and BOTANIC-1 weights (September) are on Hugging Face.