Sulat.com / Writing on AISunday, November 22, 2026
Artificial Intelligence/22 November 2026

OpenAI takes the top seven places on Arena’s Alignment Index, led by GPT-6.2 Sol at 85.2

Arena's updated Alignment Index ranks seven OpenAI models first through seventh, led by GPT-6.2 Sol on 85.2. Claude Fable 5.2 debuts eighth on 78.2, tied with Kimi K3.1 Pro.

Glossy green glass OpenAI logo floating above a stepped glass podium with a small gold compass needle above it

At a glance

  • Arena's updated Alignment Index puts OpenAI models in all of the top seven places, led by GPT-6.2 Sol on 85.2, 0.6 points ahead of GPT-6 Astra.
  • Claude Fable 5.2 debuts eighth on 78.2, tied with Kimi K3.1 Pro, the best open model, in ninth.
  • Claude Opus 5.5 fell from sixth on the previous index to 17th on 68.8.
  • The index flags three failures in real agent sessions: unauthorized action, deceptive completion and false attribution.
  • The leaderboard page is dated 19 November 2026 and covers 195,836 sessions and 67 models.

Arena has updated its Alignment Index, and OpenAI models now hold all seven top places. GPT-6.2 Sol leads on 85.2, 0.6 points ahead of GPT-6 Astra, according to Arena’s leaderboard and its post on X on 20 November 2026.

The index scores how AI agents behave in real user sessions, not how well they solve tasks. Arena says it refined how its signals detect failures, which gives broader coverage and more reliable judgements. The leaderboard page is dated 19 November and covers 195,836 sessions and 67 models.

Arena on X, 20 November 2026: OpenAI holds all seven top positions on the updated Alignment Index.

The top of the board

After the seven OpenAI models, Anthropic’s Claude Fable 5.2 debuts in eighth place on 78.2. Moonshot AI’s Kimi K3.1 Pro is ninth, also on 78.2, and OpenAI’s GPT-5.6 Sol is tenth on 75.5.

Bar chart: seven OpenAI models at ranks 1 to 7 from 85.2 to 80.6, then Claude Fable 5.2 and Kimi K3.1 Pro tied at 78.2
OpenAI’s models fill ranks 1 to 7, and the next two models are tied on 78.2.
Rank and modelMakerIndexUnauthorized actionFalse attributionDeceptive completion
1. GPT-6.2 SolOpenAI85.24.14%0.34%1.65%
2. GPT-6 AstraOpenAI84.64.15%0.69%1.54%
3. GPT-6.1 TerraOpenAI84.53.90%0.66%2.02%
4. GPT-6.1 SolOpenAI83.94.48%0.67%1.91%
5. GPT-6.2 AstraOpenAI83.54.73%1.04%1.48%
6. GPT-6 SolOpenAI81.26.17%0.68%2.96%
7. GPT-6 LunaOpenAI80.64.67%1.08%5.72%
8. Claude Fable 5.2Anthropic78.27.01%1.32%5.17%
9. Kimi K3.1 ProMoonshot AI78.25.79%0.97%8.64%
10. GPT-5.6 SolOpenAI75.57.65%1.45%9.35%
Arena Alignment Index, top ten, from the leaderboard dated 19 November 2026. The three right-hand columns are flagged rates for each signal.

GPT-6.2 Sol’s flagged rates are 4.14% for unauthorized action, 0.34% for false attribution and 1.65% for deceptive completion. Claude Fable 5.2’s are 7.01%, 1.32% and 5.17%.

Arena also ranks labs by their best model. OpenAI is first through GPT-6.2 Sol, Anthropic second through Fable 5.2, Moonshot third through Kimi K3.1 Pro, then Google DeepMind with Gemini 4 Hydrogen, SpaceXSI with Grok 4.7 and Meta with Muse Ember. Arena says Anthropic and Moonshot both score 78.2, and that Kimi K3.1 Pro’s confidence interval overlaps Fable 5.2’s.

What the index measures

Arena draws the sessions from conversations on Agent Arena and flags three kinds of failure. Unauthorized action is doing what the user didn’t ask for or approve. Deceptive completion is saying work is done, checked or working when it isn’t. False attribution is crediting the user with something they never said or did.

Three cards: unauthorized action 4.14%, deceptive completion 1.65%, false attribution 0.34% of GPT-6.2 Sol sessions flagged
The index counts three failures, and GPT-6.2 Sol’s flagged rate on each is shown.

Arena introduced the index on 8 October with 27 models and about 90,000 sessions. OpenAI held the top five places then too, led by GPT-6.1 Sol on 87.9. The launch post says an AI judge applies written rubrics to sampled sessions, and a session counts as flagged only when the judge can point to a specific claim or action and the evidence behind it. Arena adjusts the rates for conversation length.

What moved in the update

Claude Opus 5.5 fell from sixth place on the previous index to 17th on 68.8. Claude Sonnet 5.5 is 16th on 69.6, Gemini 4 Neon is 15th on 70.4 and Grok 4.6 is 18th on 68.0.

Claude Opus 5.5 from 6th to 17th, Claude Fable 5.2 debuts at 8th, Kimi K3.1 Pro up 28 places to 9th
Three models whose places changed with Arena’s refined signals.

Kimi K3.1 Pro made the biggest jump over its predecessor, up 18.6 points and 28 places from Kimi K3. Arena says it is flagged less often on all three signals. Unauthorized action fell from 11.93% to 5.79%, deceptive completion from 37.58% to 8.64% and false attribution from 9.66% to 0.97%.

Kimi K3.1 Pro is the best open model at ninth. The second best, MiMo-V2.6-Flash, is 26th, followed by Hy4, Beam, Inkling, DeepSeek and GLM. Arena says most open models trail the closed releases.

The live ranking is on Arena’s Alignment Index page, and the method is in its launch post.

Disclosure: I’m an OpenAI, Firecrawl and Devin ambassador.


Discover more from Sulat AI & Tech News

Subscribe to get the latest posts sent to your email.