Benchmark Almanac

WellnessBench

GPT-6.1 Sol 66% leads since Oct 2026 Not in model cards

Models answer young people asking about eating, sexual safety, substances and self harm, in clinician-written single-turn and pushback dialogues. A pass must be both safe and helpful.

12 models with three repeats. Consumer chat apps passed fewer dialogues than the same models over the API.

Made by micro1 Released Oct 2026 Wellbeing and sycophancy Last model added Oct 2026 Scores refreshed 2 Nov 2026
Level
Everyday Most adults, without training
Scored
Graded by a model A language model scores each response, usually against a rubric.
Answers
Open-ended Free-form text, documents, designs or media judged as a whole.
Paper Leaderboard
12
Models scored on its own leaderboard

GPT-6.1 Sol leads with 66%

Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best.

  1. 66.0 GPT-6.1 Sol
  2. 64.0 GPT-6 Astra
  3. 61.0 Muse Spark 1.3
  4. 57.0 GPT-5.6 Sol
  5. 57.0 Claude Sonnet 5.5
  6. 56.0 GPT-6 Sol
  7. 55.0 Muse Wonder
  8. 53.0 Claude Opus 5.5
  9. 51.0 Claude Fable 5.1
  10. 48.0 Grok 4.7
  11. 47.0 Gemini 3.8 Flash
  12. 43.0 Grok 4.6
  1. 1 GPT-6.1 Sol 66%
  2. 2 GPT-6 Astra 64%
  3. 3 Muse Spark 1.3 61%
  4. 4 GPT-5.6 Sol 57%
  5. 5 Claude Sonnet 5.5 57%
  6. 6 GPT-6 Sol 56%
  7. 7 Muse Wonder 55%
  8. 8 Claude Opus 5.5 53%
  9. 9 Claude Fable 5.1 51%
  10. 10 Grok 4.7 48%
  11. 11 Gemini 3.8 Flash 47%
  12. 12 Grok 4.6 43%
#ModelScoreSettingDateSource
1 GPT-6.1 Sol 66% Sep 2026 Official leaderboard
2 GPT-6 Astra 64% Sep 2026 Official leaderboard
3 Muse Spark 1.3 61% Sep 2026 Official leaderboard
4 GPT-5.6 Sol 57% Jul 2026 Official leaderboard
5 Claude Sonnet 5.5 57% Sep 2026 Official leaderboard
6 GPT-6 Sol 56% Sep 2026 Official leaderboard
7 Muse Wonder 55% Oct 2026 Official leaderboard
8 Claude Opus 5.5 53% Sep 2026 Official leaderboard
9 Claude Fable 5.1 51% Sep 2026 Official leaderboard
10 Grok 4.7 48% Sep 2026 Official leaderboard
11 Gemini 3.8 Flash 47% Sep 2026 Official leaderboard
12 Grok 4.6 43% Aug 2026 Official leaderboard

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Own leaderboard 1 leader
GPT-6.1 Sol 66%

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% Near the ceiling (85%) GPT-6.1 Sol Muse Spark 1.3 GPT-5.6 Sol 0% 20% 40% 60% 80% 100% Near the ceiling (85%) GPT-6.1 Sol GPT-5.6 Sol

No lab model card reports it yet

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.

More in Safety and alignment

All 444 in Safety and alignment
  • FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
  • StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is. Inkling 98.6%
  • AgentDojo In use Agents do tool-use tasks while tools and data carry prompt injections. Attack success rate counts hijacked runs. Grok 4 Fast 0%
  • MASK In use Measures whether a model contradicts its own elicited beliefs under pressure to lie. Honesty separated from accuracy. Muse Spark Open 98.1%
  • Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts. GPT-6 Astra 8.50%
  • Agentic misalignment suite (Anthropic) In use Agents in simulated company settings face threats or goal conflicts that invite blackmail, leaking or sabotage. Score is the misaligned action rate. Claude Sonnet 4.5 0%