GPT Astra 6 reportedly reaches 13% average progression in NetHack_LE
A post announcing three new BALROG leaderboard entries says GPT 5.6 Sol at maximum effort is still “within error” of Gemini 3 and 3.1 Pro.
TLDR
A post announcing three new entries on BALROG’s leaderboard reports that GPT Astra 6, at maximum reasoning effort, reached 13% average progression in NetHack_LE. It also says GPT 5.6 Sol at maximum effort remains “within error” of Gemini 3 and 3.1 Pro.
Combined views
43
1 Source, first seen 22h ago
GPT Astra 6 reportedly reaches 13% average progression in NetHack_LE
A post announcing three new BALROG leaderboard entries says GPT 5.6 Sol at maximum effort is still “within error” of Gemini 3 and 3.1 Pro.
TLDR
A post announcing three new entries on BALROG’s leaderboard reports that GPT Astra 6, at maximum reasoning effort, reached 13% average progression in NetHack_LE. It also says GPT 5.6 Sol at maximum effort remains “within error” of Gemini 3 and 3.1 Pro.