GPT-5.6(openai.com)
1549 points by logickkk1 63 days ago | 1101 comments
tl;dr: Summary not available
HN Discussion:
  • Highlights useful developer guidance and semantic tips from the new model documentation
  • Mocks benchmark comparisons as cherry-picked or unfair to competitors like Fable
  • Notes impressive benchmark achievements like SOTA on ARC-AGI-3
  • Criticizes the confusing naming convention of Sol/Terra/Luna variants
  • ~Mixed real-world testing results suggest improvements are less dramatic than claimed