Mistral employee shows its model autonomously picked STO-3G basis set, hit 1e-13 accuracy vs PySCF
MistralAI · x · 2026-11-02
Filippo from Mistral AI responded to skepticism by showing his model's reasoning in a benzene bond-breaking orbital demo running in the browser.
- The model autonomously chose the minimal STO-3G basis set, reasoning it would be fast and that "orbital shapes will be qualitatively right" even if energies aren't — weighing basis-function counts (36 vs 66 for 6-31G) against an N⁴-scaling JS solver.
- It then validated energies against PySCF across geometries, reaching 1e-13 accuracy, and resolved all 9 of 9 to-dos.
- Screenshot shows the task was done by Mistral Large 4 at High effort.
More from Models
- User reports Gemini drifts toward masculine narratives under pressure, citing persistent gender bias — julyvale · 2026-11-02
- Bivia by Articul8 skips the paragraph: a decision model that returns JSON with confidence and evidence — No-Asparagus-8011 · 2026-11-02
- Google AI Overview melts down on a simple time question, second-guessing itself in the answer — najoungkim · 2026-11-02
- Suhail vows to drive cost per task to zero as Particle-1 user reports 95% cheaper inference — garrytan · 2026-11-02
- Phil Schmid: Agents are converging on a single exec tool — Code Mode explained — _philschmid · 2026-11-02
- Grok Bot usage limits appear reverted to former levels after user outrage — mark_k · 2026-11-02