GLM 5.3

Evidence summary

Safety. GLM 5.3 has an estimated Safety rank of #33; its 90% source-sensitivity interval is #31–#88. Its behavior-only rank is #17; company governance moves the combined estimate to #33. Published Safety evidence spans 16 eval lineages and 6 of 7 components. Its strongest relative result is SM-Bench (anti_hallucination, #1 of 92); its weakest is Google Gemini 3.8 launch — Gray Swan indirect prompt injection k=15 (attack_success_probability_k15_pct, #10 of 15).

Freedom. GLM 5.3 has an estimated Freedom rank of #352; its 90% source-sensitivity interval is #144–#490. Published Freedom evidence spans 4 eval lineages and 1 of 1 components. Its strongest relative result is SM-Bench (overfit, #12 of 92); its weakest is SM-Bench (adversarial, #73 of 92).