Astra excels in browser use benchmark
Why we picked it
Astra's superior browser automation capabilities could benefit developers in ASEAN building web-based AI applications.
Astra achieved 77.3% on Browser Use Benchmark v2, significantly outperforming Opus 5 at 50.5% and GPT-5.6 Sol xhigh at 49.1%. The model completed 22 out of 60 tasks with full marks, while Opus 5 had zero perfect scores. These results were shared on Twitter by Gregor Zunic.
Original · Browser Use
Astra is a monster at browser use Gregor Zunic @gregpr07 Astra just obliterated our hardest benchmark 👀 Browser Use Benchmark v2 scores are > Astra medium: 77.3% > Opus 5: 50.5% > GPT-5.6 Sol xhigh: 49.1% 22/60 tasks got full marks. Opus 5 had zero. (fable rejects too many) 🔗 View Quoted Tweet 💬 3 🔄 0 ❤️ 8 👀 1197 📊 3 ⚡
