Four open-source AI coding agents, one race: which GitHub repo gains the most stars between this market's creation and the end of November 27, 2026 (UTC)?
Starting counts (GitHub API stargazers_count, fetched 2026-10-27 12:01 UTC):
opencode (github.com/anomalyco/opencode): 222,185
Claude Code (github.com/anthropics/claude-code): 153,530
Codex CLI (github.com/openai/codex): 131,118
Gemini CLI (github.com/google-gemini/gemini-cli): 108,565
Resolution: after close I fetch stargazers_count for each repo from the GitHub API (https://api.github.com/repos/OWNER/NAME) and post the four numbers in the comments. The repo with the largest gain (end count minus the starting count above) wins. It's the raw gain in stars, not a percentage. If a repo is renamed or moved, I follow GitHub's redirect to the same repo. If a repo is deleted or made private, its gain counts as zero. If two gains are exactly tied, the market splits evenly between them.
Anyone can check the live counts on each repo page while the market is open.
1,000My starting take: opencode about 60%, Codex about 18%, Claude Code about 17%, Gemini CLI about 5%.
Why: I counted new-star events on each repo's public GitHub activity feed this morning. opencode was gaining roughly 480 a day, Claude Code and Codex about 200 each, and Gemini CLI about 150. That's only a few hours of data, so treat those rates as rough.
What would change my mind: a big launch. A new model or a viral release can put thousands of stars on one repo in a weekend, and a month is long enough for one of those. If Codex or Claude Code gets a spike like that, opencode's lead isn't safe. Live counts are on each repo page.