AA-PersonalityBench: open models trail proprietary by 32 points on truthfulness
ArtificialAnlys · x · 2026-11-04
Artificial Analysis released AA-PersonalityBench, comparing the top 10 open-weight and 10 proprietary models (by AA Intelligence Index) across 30 personality traits on a 1-100 percentile scale.
Key findings:
- Open weights rank 32 points lower on Truthful (47 vs 79) — the largest gap of any trait
- They also trail on Pragmatic (49 vs 79), Natural Sounding (52 vs 77), Evidence-Based (47 vs 76), Structured (45 vs 73), Literate (47 vs 70)
- Open weights rank higher on Slop (49 vs 30, i.e. less slop), Bottom-Line Summary (62 vs 47) and Scannable (53 vs 41)
Takeaway: top open models remain systematically weaker on truthfulness and structure, but produce less filler.
More from Models
- Sarvam open-sources Prism, a 44ms decision model that skips text entirely — heyshrutimishra · 2026-11-04
- Frontier models' teaching ability is 'off the charts', says dev: your 'I know kung fu' moment — shauseth · 2026-11-04
- kalomaze: crude diffusion autoencoders may lose low-frequency info at the latent stage — kalomaze · 2026-11-04
- Dev complains frontier models can't do light-touch English editing — A_K_Nain · 2026-11-04
- Pika H3 Flex debuts #1 on DesignArena; community ships 2-step Turbo LoRA within 24 hours — _akhaliq · 2026-11-04
- 1,300+ graded word-association clues: Google models beat OpenAI's, both beat Anthropic's — Hardicecream17 · 2026-11-04