// Tag
We tested open-weight models across coding, reasoning and tool-use tasks. Results, cost and what the numbers mean.
Oct 14, 2026 · 6 min read · 👁 –