Model benchmarking
Methods and disputes over measuring AI model performance. Every AI Coding Weekly story on model benchmarking, newest first. 3 stories since October 2026.
- Theo gave three models one bug, Sol's fix did not work
Opus filed a PR, Muse pointed at someone else's existing patch, and Sol crashed on its own question tool before shipping a fix that failed.
- Anthropic ships Haiku 5.5 at about 75% less to run
Boris Cherny calls it a good Haiku and puts the gap at 10x, Theo found out from a screenshot.
Have the good opinion, not the loud one
One email a day. Under three minutes. What working engineers found out about coding with AI.
- BridgeMind turns down Theo's $10,000 NerfBench bet
The fight over whether anyone can measure model nerfing ends with no transcripts and no bet.
People on Model benchmarking
Get the next one by email
Coding with AI, read daily so you do not have to. The experiments, the receipts and the arguments from engineers shipping real software. Not a changelog.