Search for a command to run...
In the previous articles I covered the SWE-Bench numbers, the open-weight models worth knowing, and why bigger context doesn't mean better output. Now I want to connect everything with two things I us
When a company says their model has 1 million tokens of context, everyone assumes bigger is better. I thought the same thing. But after running LLMs in production for a while, I found out it's not tha
In the last article I showed the SWE-Bench numbers. Open-weight models are basically tied with the proprietary ones now. Two models stood out to me: MiniMax M2.5 and Qwen3.5. Here's what I found out a
I was looking at the SWE-Bench Verified leaderboard last week and the numbers surprised me. The gap between proprietary and open-weight models is almost gone. Not in some academic test. In actual bug
