deterministic trace
Move-by-move replay
seed 101GPT-5.6 Terra #4gpt-5.6-terra
♜8
♞
♝
♛
♚
♝
♞
♜
♟7
♟
♟
♟
♟
♟
♟
♟
6
5
4
3
♙2
♙
♙
♙
♙
♙
♙
♙
♖1a
♘b
♗c
♕d
♔e
♗f
♘g
♖h
Pi / GPT-5.6 Luna #4gpt-5.6-luna
ply 0/80initial position
plysidemoveruntime
match inputs- position
- initial-position
- seed
- 101
- max plies
- 80
- result SHA
- 7a00c77b1748fa140d8889b7
competing artifacts- white SHA
- 8ed212384df2ffd09a15c642
- black SHA
- 4cf6784494c360e88fb44641
local reproductionDownload the pinned Hugging Face release, then verify its hashes, row counts, compressed bundles, and replays.
npm run verify:hf-results -- --output <downloaded-release-root>