Been using SWE 1.7 a lot in @DevinAI and it’s getting hard to tell the difference between it and Opus 4.8 or gpt 5.5. It’s also free right now, so that’s nice. Definitely reaching for Fable or gpt 5.6 on harder stuff but a huge chunk of engineering can be done now for drastically lower prices. Pretty sure in 6-12 months the majority of engineering is going to be done in non-frontier lab harnesses (Cursor, Devin, Amp, Factory) with models they’ve trained or fine-tuned. Also with the downward…