Loading…
The reported misses highlight limits in these models’ verification performance, but the unpublished test set and incomplete report leave the result provisional.
Limits of the evidence: Single press report; the designs are unpublished, results are attributed to Kumar’s independent work, and the supplied text is truncated before the GLM-5.2 result or further methodology and outcomes.
What to watch
Assessment revised Nov 2, 2026
Putting AI To The Test
Semiconductor EngineeringNov 2, 2026First report
First assessment, from Semiconductor Engineering
No maturity ladder applies to this desk.
Sens.ai aggregates and assesses published reporting. The assessment above is machine generated; the original sources are authoritative.