OpenScience agent outscores Codex and Claude Code on science benchmarks
An Apache 2.0 agent beating proprietary coding agents on science benchmarks changes the calculus for labs that cannot ship data to closed platforms. Teams running research automation can self-host it and plug in whichever model they already pay for. The comparison set — Codex and Claude Code — makes the leaderboard claims directly checkable against tools practitioners use daily.
Synthetic Sciences launched OpenScience, an open-source agent for scientific research workflows released under the Apache 2.
0 licence.On Terminal-Bench-Science, a set of 70 research workflows hosted by Stanford and the Laude Institute, OpenScience solved 53 tasks for a 75.
7% score.That beats Codex with GPT-6 Astra, which scored 68.
1% as the top entry on a 23 September mirror of the public leaderboard.On the 14 science tasks of Terminal-Bench 4.
0, OpenScience reached 71.4% versus 60.
0% for Claude Code on Claude Fable 5.1.
The workbench supports any provider's model.
Synthetic Sciences just launched its open-source OpenScience agent now outscores Codex and Claude Code on agentic science benchmarks.
A free Apache 2.0 workbench for any provider's model.
On Terminal-Bench-Science, 70 research workflows hosted by Stanford and the Laude Institute, OpenScience solved 53 tasks for 75.7%.
Codex with GPT-6 Astra scored 68.1%, the top entry on a September 23 mirror of the public leaderboard.
The gap widened on Terminal-Bench 4.0's 14 science tasks, where OpenScience scored 71.4% and Claude Code on Claude Fable 5.1 reached 60.0%.
GitHub link in comment
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include OpenScience, Synthetic Sciences, Codex, Claude Code.
Source attribution: curated from rohanpaul_ai (@rohanpaul_ai). Cross-source corroboration: 11 independent sources have covered this development (including arXiv cs.AI, Artificial Analysis, Tw93), confirming industry-wide signal strength.
