Reaction
Mistral Large 4 gets a favorable STEMBench assessment
A tester says the model got a "pretty nice score" on 20 STEMBench problems in each of six domains.
TLDR
A tester says Mistral Large 4 got a "pretty nice score" on their STEMBench test and performed fairly evenly across six domains. They argued it is capable of original research across domains. Another poster called a separate quoted post "absolutely brutal for Mistral," then pointed to the favorable test "on the other hand."
Combined views
625
1 Source, first seen ago
