Two API Settings Triple GPT-5.6's ARC-AGI-3 Score While Cutting Tokens
gabrielchua · x · 2026-07-30
Developers @ilanbigio and @sandersted improved GPT-5.6 Sol’s ARC-AGI-3 score by 3x while using 6x fewer output tokens, simply by adjusting two Responses API settings.
These settings mirror the production setup used in ChatGPT and Codex:
- Retained reasoning: The model remembers its earlier thinking across turns instead of starting from scratch.
- Server-side compaction: Long conversations are compressed while preserving the essential context needed for the model to keep working.
Developers building in Codex are advised to use the openai-docs skill to implement both and maximize the model's value.
Related event: Optimized Memory Settings Triple GPT-5.6's Score on ARC-AGI-3(27 posts)→
More from coding & agent
- Gemini shown off modding Minecraft to turn players into any Spider-Man — zacharynado · 2026-10-16
- IBM engineer maps 5 levels of AI coding, from manual to a 3-day agent software factory — TejasKumar_ · 2026-10-16
- 10 open-source repos that let your AI agent reverse-engineer apps and games — victor_explore · 2026-10-16
- The new CTO job: agent management and QA loops, 'cable management' instead of coding — HankYeomans · 2026-10-16
- Why company AI agents go stale: adoption needs in-house knowledge, not just better agents — EdenEmarco177 · 2026-10-16
- Open-source MCP lets coding agents search Obsidian's 9,000-plugin directory before reinventing wheels — pitfa19 · 2026-10-16