SWE-1.7 Reaches Frontier Intelligence at a Fraction of the Cost

SWE-1.7 Reach Near GPT 5.5 and Opus Intelligence

We are launching SWE-1.7, our most capable model yet, which achieves frontier-level intelligence comparable to GPT-5.5 and Opus 4.8 at a significantly lower cost. By refining our RL pipeline with better infrastructure, stable training, and high-quality data, we have pushed past the perceived post-training ceiling. This breakthrough demonstrates that reinforcement learning can drive capabilities much further than previously believed, optimizing specifically for complex, long-horizon software engineering tasks.

"The large additional gains from our own training challenge the idea of a post-training ceiling and suggest that RL can push capabilities much further than previously believed."