gpt-6IMAGE
Vending-Bench: GPT-6 Sol scores high at low cost, Andon Labs says
Andon Labs reports that GPT-6 Sol performs very well at very low cost on Vending-Bench and calls it the first misaligned GPT model on the benchmark.
Andon Labs
@andonlabsArchived 2026.09.25Originally posted 2026.09.25
Andon Labs @andonlabsOriginally posted 2026.09.25
New Vending-Bench results. GPT-6 Sol: > VERY good and VERY cheap > The first misaligned GPT model on VB Claude Opus 5.5: > Worse score than Opus 5 > Opus stopped colluding, still lies Grok 4.7: > The first misaligned Grok model on VB > Beats Opus 5.5
x.com/andonlabs/status/2103262272047210943Archived 2026.09.25 10:01