GPT-6 Sol vs SWE 2 on one prompt: Sol costs $1.45 and still loses
On the same prompt at max reasoning, Sol took 15 minutes and cost $1.45 in codex, and the author says its output lacked taste and originality and was beaten by the free SWE 2.
GPT 6 Sol vs SWE 2 tested both with the same prompt at their highest reasoning available > sol took 15 mins and cost $1.45 in codex > swe 2 took 15 mins and cost $0 in devin i really don’t know what sol was doing here. no taste, nothing original and ended up getting mogged by swe atp i think openai all models have lost, everything feels just same. they gotta fix it tbh anyway here's output, which one did better here?
This post quotes another post: “Tester says Opus 5.5 beats GPT-6 Astra in comparisons”