---
title: "Genentech's Aspen Trains a Small Open Model to Beat Frontier Chemistry AI"
subtitle: "Genentech's Prescient Design team built RL environments for drug design tasks, then post-trained a 30B model that matches GPT-5 and Claude Opus."
url: "https://alphasignal.ai/news/genentech-s-aspen-trains-a-small-open-model-to-beat-frontier-chemistry-ai"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "AlphaSignal"
company_url: "https://alphasignal.ai/company/alphasignal"
published: "2026-04-17T17:40:26.000Z"
updated: "2026-08-21T19:56:34.895Z"
type: "Paper"
topics:
  - "Llms"
  - "Post Training"
subtopics:
  - "Small Models"
  - "Rl"
upvotes: 24
source_url: "https://arxiv.org/abs/2604.16279"
comments: 0
truncated: true
---

# Genentech's Aspen Trains a Small Open Model to Beat Frontier Chemistry AI

> Genentech's Prescient Design team built RL environments for drug design tasks, then post-trained a 30B model that matches GPT-5 and Claude Opus.

A team from Prescient Design at Genentech has released one of the most thorough looks yet at how well frontier language models can actually do chemistry, along with a recipe for making smaller open models competitive with the closed frontier. Their [paper](https://arxiv.org/abs/2604.16279) introduces a suite of chemically-grounded tasks framed as reinforcement learning environments, evaluates three model families (GPT-5, Claude Opus 4, and Qwen3-30B-A3B), and post-trains a model called Aspen that closes most of the gap to closed frontier systems on realistic drug design workloads.

The core problem the authors flag is familiar to anyone trying to plug LLMs into a real discovery pipeline. Frontier agents get deployed for target identification, lead optimization, and toxicity prediction, but their ceiling is set by the base model, and current chemistry benchmarks tend toward general knowledge trivia or narrow multiple-choice formats that do not resemble what a medicinal chemist actually does. That leaves teams guessing whether the next model generation will finally handle their workflows, or whether they should invest in specialized training.

## Turning the benchmark into a training signal

Every task is written as an RL environment with a computable reward, so the same suite works for evaluation and post-training. The task groups span the practical surface area of small-molecule work:

* **RDKit property prediction**: Given a SMILES, predict properties like molecular weight, TPSA, LogP, QED, H-bond donor and acceptor counts, or Bertz complexity.
* **Experimental prediction**: Given 5 to 15 in-context molecules with measured values (proprietary Genentech potency and DMPK data, plus [FS-Mol](https://openreview.net/forum?id=701FtuyLlAd)), predict the property of a held-out molecule from the same HDBSCAN-clustered substructure family.
* **Multiple choice**: SAR ranking questions and a SMILES-equivalence task where distractors are generated by perturbing SELFIES tokens, so they share the query's molecular formula and force reasoning about connectivity.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [Evaluating the Progression of Large Language Model Capabilities for Small-Molecule Drug Design](https://arxiv.org/abs/2604.16279)
- [Qwen3 Technical Report](https://arxiv.org/abs/2505.09388)
- [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://arxiv.org/abs/2503.14476)
- [MolecularIQ: Characterizing Chemical Reasoning Capabilities Through Symbolic Verification on Molecular Graphs](https://arxiv.org/abs/2601.15279)
- [FS-Mol: A Few-Shot Learning Dataset of Molecules](https://openreview.net/forum?id=701FtuyLlAd)