1.9B 'decision-making' model claims to beat GPT-5.6 in satirical take on AI benchmarks

matlabulous · x · 2026-09-21

FLock's satirical "THIS / THAT Model 1.0" claims to be a 1.9B-parameter open-source model built purely for making decisions, posting a 94.1% success rate on a 68-question decision test and "outperforming" GPT-5.6, Deepseek-4.1-flash, Kimi K3 and Jev at 31ms latency on an RTX 5080 laptop. The fictional comparison models make it a parody of benchmark-chasing model-launch marketing.

Original post →

More from Fun

Fun channel →