Humanity's Last Exam

General

Expert-level questions across 50+ subjects from 1000+ experts, published in Nature

Source ↗ Higher is better

Rank	Model	Provider	Score
1	Claude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)	Anthropic	61.4%
2	Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Anthropic	59.1%
3	Claude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Anthropic	58.7%
4	Claude Opus 5.5 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Anthropic	57.5%
5	Gemini 4 Argon (High)	Google	57.1%
6	Claude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Anthropic	55.9%
7	Claude Opus 5.5 (Adaptive Reasoning, High Effort, Default Fallback)	Anthropic	55.6%
8	Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)	Anthropic	55.5%
9	Claude Sonnet 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)	Anthropic	55.0%
10	Claude Opus 5 (Adaptive Reasoning, Max Effort)	Anthropic	54.9%
11	Claude Opus 5.5 (Adaptive Reasoning, Medium Effort, Default Fallback)	Anthropic	54.7%
12	GPT-6 Astra (max)	OpenAI	54.7%
13	GPT-6 Astra (xhigh)	OpenAI	54.6%
14	Claude Opus 5 (Adaptive Reasoning, Xhigh Effort)	Anthropic	54.4%
15	Claude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Anthropic	53.8%
16	Muse Wonder Full (Max)	Meta	53.6%
17	GPT-6 Astra (high)	OpenAI	53.1%
18	GPT-6.1 Sol (max)	OpenAI	52.9%
19	Claude Opus 5 (Adaptive Reasoning, High Effort)	Anthropic	52.8%
20	GPT-6 Astra (medium)	OpenAI	52.7%
21	GPT-6.1 Sol (xhigh)	OpenAI	52.6%
22	GPT-6.1 Sol (high)	OpenAI	51.4%
23	Claude Opus 5 (Adaptive Reasoning, Medium Effort)	Anthropic	51.3%
24	Muse Wonder (Max)	Meta	50.1%
25	Claude Sonnet 5.5 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Anthropic	50.0%
26	GPT-6.1 Sol (medium)	OpenAI	49.9%
27	GPT-5.6 Sol (max)	OpenAI	49.5%
28	Xiaomi: MiMo-V2.6-Pro	Xiaomi	49.4%
29	GPT-6 Astra (low)	OpenAI	49.2%
30	Claude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Anthropic	48.9%
31	Claude Opus 4.8 (Adaptive Reasoning, Max Effort)	Anthropic	48.7%
32	Muse Spark 1.3 (max)	Meta	48.7%
33	Claude Opus 5.5 (Adaptive Reasoning, Low Effort, Default Fallback)	Anthropic	48.3%
34	Gemini 3.7 Flash (high)	Google	47.9%
35	GPT-6 Sol (max)	OpenAI	47.9%
36	Gemini 3.8 Flash (high)	Google	47.8%
37	Muse Spark 1.3 (xhigh)	Meta	47.5%
38	GPT-6.1 Sol (low)	OpenAI	47.4%
39	GPT-5.6 Sol (xhigh)	OpenAI	47.3%
40	Google: Gemini 3.1 Pro Preview	Google	47.0%
41	Kimi K3 (max)	Moonshot AI	46.9%
42	StepFun: Step 5 Preview	Stepfun	46.5%
43	GPT-6 Sol (xhigh)	OpenAI	46.3%
44	Muse Spark 1.1 (xhigh)	Meta	46.2%
45	GPT-5.6 Sol (high)	OpenAI	46.0%
46	GPT-5.5 (xhigh)	OpenAI	45.8%
47	Claude Sonnet 5.5 (Adaptive Reasoning, High Effort, Default Fallback)	Anthropic	45.8%
48	Muse Spark 1.2 (xhigh)	Meta	45.5%
49	GPT-5.5 (high)	OpenAI	45.0%
50	Claude Haiku 5.5 (Max)	Anthropic	44.4%

Access this data

JSON GET /api/v1

[
  {
    "rank": 1,
    "model": "Claude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)",
    "provider": "Anthropic",
    "score": "61.4"
  },
  {
    "rank": 2,
    "model": "Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)",
    "provider": "Anthropic",
    "score": "59.1"
  },
  {
    "rank": 3,
    "model": "Claude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)",
    "provider": "Anthropic",
    "score": "58.7"
  }
]

Curl

$ curl -s https://rightrank.ai/api/v1/rankings/humanity-s-last-exam.json | jq '.[:3]'

CLI brew install rightrank

$ rightrank top humanity-s-last-exam --limit 5
$ rightrank compare claude-opus-4.6 gpt-5.4-pro --category humanity-s-last-exam
$ rightrank value humanity-s-last-exam  # best performance per dollar

Claude Code Skill Skill

# In Claude Code, use the RightRank skill:
/rightrank best model for humanity-s-last-exam
/rightrank compare top 5 by value
/rightrank cheapest model scoring above 80 on humanity-s-last-exam