Artificial Analysis launches AA-PersonalityBench with custom rankings
11/04/2026 — 11/06, 01:20·1 sources (all)·7 reports (all)
Key takeaways
- Artificial Analysis has launched AA-PersonalityBench, a page that lets users build custom model rankings across personality dimensions, insp…
Background & analysis
On 11/4/2026, the AI benchmarking firm Artificial Analysis launched a new evaluation page called AA-PersonalityBench and published a set of model comparisons alongside it. The page lets users build their own personality rankings, dig into detailed test results, and read the published methodology. Artificial Analysis framed the tool as a way for developers and users to get a fuller picture of how different large models behave in terms of personality and persona.
AA-PersonalityBench compares 10 top open weights models against 10 top proprietary models on a set of personality indexes. Looking at average index scores, the proprietary group holds the lead on most dimensions, scoring 68 versus 45 on Informativeness and ranking higher on average on truthfulness, pragmatism, and structure. The open weights models come out ahead only on slop, where they average 49 against 30, and they also perform relatively well at summarizing the bottom line. Artificial Analysis stresses that these are group averages, and individual models vary within each group.
The results sit under Artificial Analysis's broader Intelligence Index. The firm published the ten highest scoring models in the proprietary group on that index, with Claude Opus 5.5, GPT-6 Astra, GPT-6.1 Sol, Claude Sonnet 5.5 (max), and Gemini 4 among the current flagship systems included. As of publication, this round of comparisons and rankings is the latest available, and the AA-PersonalityBench page is open for users to build their own personalized rankings. No further updates have been released.
One caveat worth keeping in mind: the published results are group averages, so individual models within each group can differ noticeably from the headline numbers, and the comparison covers only the 10 open weights and 10 proprietary models that Artificial Analysis measured.
AI-generated from 4 reports · updated 2 days ago
Latest turnArtificial Analysis has opened an AA-PersonalityBench page where users can build their own personality rankings and dig into more detailed per-model results. The company also published the benchmark methodology alongside it.
- Heat index
- 91
- All sources
- 1
- All reports
- 7
- First seen
- 2 days ago
Reports on this story headlines open the original
Artificial Analysis has opened an AA-PersonalityBench page where users can build their own personality rankings and dig into more detailed per-model results. The company also published the benchmark methodology alongside it.
X：Artificial Analysis (@artificialanlys)First-partyAI score 45In Artificial Analysis's AA-PersonalityBench, MiMo-V2.6-Pro ends 34% of its responses with a question, the highest share of any model measured. Across the models tested, the share ranges from 34% down to 0%, with six models falling below 0.5%.
X：Artificial Analysis (@artificialanlys)First-partyAI score 43
Artificial Analysis is letting users build their own model personality rankings and explore more detailed results on the AA-PersonalityBench page. The organization also published the methodology behind the benchmark.
X：Artificial Analysis (@artificialanlys)First-partyAI score 47
Artificial Analysis has released the AA-PersonalityBench page, allowing users to build custom personality rankings and explore detailed model evaluations and methodologies.
X：Artificial Analysis (@artificialanlys)First-partyAI score 61Artificial Analysis has highlighted the ten highest-scoring proprietary models on its Intelligence Index, featuring systems like Claude Opus 5.5, GPT-6 Astra, and Gemini 4.
X：Artificial Analysis (@artificialanlys)First-partyAI score 60Artificial Analysis data shows that top open weights models generally rank lower than proprietary alternatives across personality indexes, except in metrics like slop where they score higher.
X：Artificial Analysis (@artificialanlys)First-partyAI score 60Results from AA-PersonalityBench indicate that top open weights models trail proprietary models in truthfulness, pragmatism, and structure, while scoring higher in summarizing the bottom line.
X：Artificial Analysis (@artificialanlys)First-partyAI score 60
Other stories people are talking about
- Heat index 47Artificial Analysis Launches New API Performance Benchmark Centered on Time per Task1 sources
- Heat index 642RisingOpenAI GPT-5.6 Brings Major Agentic and Computer Use Upgrades13 sources
- Heat index 840OpenAI ships Images 2.6 with GIF and SVG output15 sources
- Heat index 675OpenAI says its cyber model pipeline confirmed 1,724 Linux kernel flaws and generated 948 privilege-escalation exploits11 sources
- Heat index 633RisingCat Wu, product lead for Claude Code, joins OpenAI9 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion