aa-opencode: OpenCode models with benchmarks and price in one table

Posted on Wed, 21 Oct 2026 in Tools • Tagged with opencode, ai, benchmarks, pricing, cli, python

aa-opencode crosses the OpenCode model catalog with Artificial Analysis benchmarks and prints one table with intelligence, speed, and price per model. A Python CLI I just published.


Continue reading

HarnessTax: The Harness Matters Less Than You Think

Posted on Thu, 17 Sep 2026 in ia • Tagged with pi, opencode, coding-agent, ai, harness, claude-code, codex, benchmarks, ai-agents

The HarnessTax study from UC Berkeley and Arena evaluates 21 model-harness pairings and reaches three surprising conclusions. They back up with data why I use pi and OpenCode daily.


Continue reading