π I am seeing errors when using opencode against my GitHub Copilot Business subscription with the GPT-4.1 model (for miserly reasons), indicating that the context window had been filled before opencode had decided to autocompact.
The error message seen in opencode seems pretty unambiguous, clearly stating that the allowed prompt length is 64000 tokens.
prompt token count of 64314 exceeds the limit of 64000
To try to corroborate this, I have also used GitHub's Copilot CLI tool, which clearly states that the context size for the GPT-4.1 model is 64k:
I have been unable to find any authoritative source listing the context sizes though.