It was a bit of a silly challenge, wasn’t sure how workable, learned a lot in the process about what actually drives usage / costs, and how to keep both under control
Flash is pretty decent coder, but it should be paired with good planner and reviewer. I would pick astra low for planning and sol 6.1 medium for reviews.
What would you use if you wanted to stay (at least) open weight?
Deepseek 4.1 Flash and Mimo 2.6 Flash.
qwen3.8, kimi3, kimi2.7, GLM-5.3 are all good families I use in my coding team
I'm mainly using flash varients, at least as the default, bump.up to stronger model as needed (less often these days)
Flash is plenty good for planning and reviewing, for my needs. In fact, I use it for that because it's too slow for execution, despite the name (from z.ai).
Considering they were your top two models, how did the flash and non-flash versions compare? Did you use them for different tasks?
Worth a comparison with DeepSeek v4.1 flash, if you've got another month to spare!
They've got an image of their homegrown benchmark in the post that lists DS4.1.