Benchmark leaderboard

Humanity's Last Exam (Artificial Analysis run)

Humanity's Last Exam is a closed-ended set of expert-written questions across dozens of academic fields, built to stay hard after older exams saturated. Artificial Analysis runs it independently with the same prompting and grading for every model. This page lists every model ModelCap tracks with a published AA Humanity's Last Exam result, its best evaluated configuration, the source's own rank, and the model's ModelCap Index position, API price and context window.

Snapshot as of 16 October 2026

Claude Opus 5.5 from Anthropic leads the Humanity's Last Exam (Artificial Analysis run) among the 87 tracked models with 61.4% (Reasoning configuration), per the source snapshot published 16 October 2026.

Tracked models with a result

Entries on the source board

267

Open-weight models listed

AA Humanity's Last Exam standings (87 models)

Sorted by published score, one row per model. Source: Artificial Analysis.

Humanity's Last Exam (Artificial Analysis run): every tracked model's published score, source rank, ModelCap Index position, price and context
#	Model	Score	Configuration	Source rank	ModelCap Index	Output $/1M	Context	Weights	Published
1	Claude Opus 5.5Anthropic	61.4%	ReasoningSource label: Max, Default FallbackNormalized effort: Unknown	#1 / 267	#2	$20.00	1M	No linked weights	6 October 2026
2	Claude Fable 5.1Anthropic	59.1%	ReasoningSource label: Max, Default FallbackNormalized effort: Unknown	#2 / 267	#3	$50.00	1M	No linked weights	6 October 2026
3	Gemini 4 ArgonGoogle	57.1%	HighSource label: HighNormalized effort: high	#5 / 267	#1	$10.00	1M	No linked weights	13 October 2026
4	Claude Sonnet 5.5Anthropic	55.0%	ReasoningSource label: Max, Default FallbackNormalized effort: Unknown	#8 / 267	#15	$10.00	1M	No linked weights	6 October 2026
5	GPT-6 AstraOpenAI	54.7%	MaxSource label: MaxNormalized effort: max	#10 / 267	#11	$50.00	1M	No linked weights	30 September 2026
6	GPT-6.1 SolOpenAI	52.9%	MaxSource label: MaxNormalized effort: max	#15 / 267	#8	$10.00	1M	No linked weights	30 September 2026
7	Muse WonderMeta	50.1%	MaxSource label: MaxNormalized effort: max	#19 / 267	#4	$20.00	1M	No linked weights	12 October 2026
8	MiMo-V2.6-ProXiaomi	49.4%	Reasoning	#22 / 267	#18	$0.87	1M	Open weights	21 September 2026
9	Muse Spark 1.3Meta	48.7%	MaxSource label: MaxNormalized effort: max	#25 / 267	#6	$4.25	1M	No linked weights	30 September 2026
10	Gemini 3.8 FlashGoogle	47.8%	HighSource label: HighNormalized effort: high	#27 / 267	#5	$4.13	1M	No linked weights	30 September 2026
11	Gemini 3.1 Pro PreviewGoogle	47.0%	Reasoning	#30 / 267	#10	$12.00	1M	No linked weights	5 September 2026
12	Kimi K3Moonshot AI	46.9%	MaxSource label: MaxNormalized effort: max	#31 / 267	#12	$11.94	1M	Restricted license	30 September 2026
13	Step 5 PreviewStepFun	46.5%	Reasoning	#32 / 267	#42	$2.70	1M	Restricted license	8 October 2026
14	Claude Haiku 5.5Anthropic	44.4%	MaxSource label: MaxNormalized effort: max	#34 / 267	#58	$0.50	1M	No linked weights	8 October 2026
15	Muse Spark OpenMeta	43.9%	X-HighSource label: XhighNormalized effort: xhigh	#35 / 267	#16	$4.25	1M	Gated access	15 October 2026
16	Grok 4.7xAI	43.1%	X-HighSource label: XhighNormalized effort: xhigh	#37 / 267	#46	$6.00	500K	No linked weights	30 September 2026
17	Qwen3.8 Max (0902)Qwen	43.1%	ReasoningSource label: 0902Normalized effort: Unknown	#38 / 267	#34	$6.00	1M	No linked weights	15 September 2026
18	GPT-5.6 TerraOpenAI	42.9%	MaxSource label: MaxNormalized effort: max	#39 / 267	#28	$12.00	1M	No linked weights	30 September 2026
19	GPT-5.3-CodexOpenAI	42.5%	X-HighSource label: XhighNormalized effort: xhigh	#41 / 267	—	$14.00	400K	No linked weights	30 September 2026
20	GLM 5.3Z.ai	42.3%	MaxSource label: MaxNormalized effort: max	#44 / 267	#20	$1.80	1M	Restricted license	30 September 2026
21	DeepSeek V4 Pro 0813DeepSeek	41.0%	MaxSource label: MaxNormalized effort: max	#47 / 267	#29	$1.98	1M	Open weights	30 September 2026
22	GLM 5.3 FlashZ.ai	39.9%	Reasoning	#48 / 267	#23	$0.20	1M	Open weights	8 October 2026
23	Ling 3.1 FlashInclusionAI	39.4%	Reasoning	#50 / 267	#35	$0.225	262K	Open weights	6 October 2026
24	DeepSeek V4.1 FlashDeepSeek	39.2%	MaxSource label: MaxNormalized effort: max	#51 / 267	—	$0.144	1M	Open weights	30 September 2026
25	MiniMax M3MiniMax	39.0%	Reasoning	#53 / 267	#48	$1.20	1M	Restricted license	5 September 2026
26	GPT-6 LunaOpenAI	38.5%	MaxSource label: MaxNormalized effort: max	#54 / 267	#45	$0.50	1M	No linked weights	30 September 2026
27	Qwen3.8 FlashQwen	38.0%	Reasoning	#56 / 267	#89	$0.47	1M	Restricted license	5 September 2026
28	Qwen3.7 PlusQwen	35.6%	Reasoning	#63 / 267	#36	$1.28	1M	No linked weights	5 September 2026
29	MiMo-V2.6-FlashXiaomi	35.1%	Reasoning	#64 / 267	#39	$0.28	1M	Open weights	27 September 2026
30	Kimi K2.7 CodeMoonshot AI	35.0%	Reasoning	#65 / 267	#59	$3.35	262K	Restricted license	5 September 2026
31	Qwen3.8 27BQwen	33.9%	X-HighSource label: XhighNormalized effort: xhigh	#71 / 267	#57	$2.20	1M	Open weights	30 September 2026
32	Nex-N2-ProNex AGI	33.7%	ReasoningSource label: Based on Qwen3.5-397B-A17BNormalized effort: Unknown	#73 / 267	—	—	262K	Open weights	30 September 2026
33	Hy3Tencent	33.5%	Reasoning	#75 / 267	—	$0.528	262K	Open weights	5 September 2026
34	Inkling SmallThinking Machines	33.3%	Reasoning	#76 / 267	#100	$1.20	524K	Open weights	5 September 2026
35	InklingThinking Machines	31.9%	X-HighSource label: XhighNormalized effort: xhigh	#80 / 267	#52	$4.05	524K	Open weights	30 September 2026
36	Solar Pro 4Upstage	29.8%	Reasoning	#82 / 267	#119	$0.36	524K	No linked weights	16 October 2026
37	Qwen3.5 397B A17BQwen	29.0%	ReasoningSource label: ReasoningNormalized effort: Unknown	#85 / 267	#44	$3.60	262K	Open weights	13 September 2026
38	Solar Open2 250BUpstage	28.5%	Reasoning	#86 / 267	#53	—	—	Restricted license	9 September 2026
39	Solar Mini 4Upstage	25.8%	Reasoning	#90 / 267	#51	$0.20	524K	No linked weights	1 October 2026
40	Qwen3.5-122B-A10BQwen	25.2%	ReasoningSource label: ReasoningNormalized effort: Unknown	#91 / 267	#81	$2.08	262K	Open weights	13 September 2026
41	Ling 3.0 Flash FinInclusionAI	22.6%	Reasoning	#96 / 267	#111	—	262K	Open weights	17 September 2026
42	Qwen3.6 35B A3BQwen	22.2%	ReasoningSource label: ReasoningNormalized effort: Unknown	#97 / 267	#95	$1.00	262K	Open weights	13 September 2026
43	Ling 3.0 Flash VLInclusionAI	22.0%	Reasoning	#98 / 267	#50	$0.18	131K	Open weights	12 September 2026
44	Ring 2.6 1TInclusionAI	21.6%	Reasoning	#100 / 267	#87	—	—	Open weights	9 September 2026
45	o3OpenAI	20.1%	Reasoning	#102 / 267	#67	$8.00	200K	No linked weights	5 September 2026
46	gpt-oss-120bOpenAI	19.6%	HighSource label: HighNormalized effort: high	#105 / 267	#156	$0.60	131K	Open weights	30 September 2026
47	Gemini 3.5 Flash LiteGoogle	18.8%	Reasoning	#107 / 267	#37	$2.50	1M	No linked weights	5 September 2026
48	Trinity Large ThinkingArcee AI	15.8%	Reasoning	#116 / 267	#133	$0.80	262K	Restricted license	9 September 2026
49	Qwen3.5-9BQwen	14.9%	ReasoningSource label: ReasoningNormalized effort: Unknown	#118 / 267	#134	$0.15	262K	Open weights	13 September 2026
50	Mistral Medium 3.5Mistral AI	13.8%	Reasoning	#125 / 267	#74	$7.50	262K	No linked weights	5 September 2026
51	Nemotron Cascade 2 30B A3BNVIDIA	12.0%	Reasoning	#131 / 267	#182	—	—	Restricted license	9 September 2026
52	Command A+Cohere	12.0%	Reasoning	#133 / 267	#176	$1.50	192K	No linked weights	23 September 2026
53	Gemma 4 31BGoogle	11.8%	Non-reasoningSource label: Non-reasoningNormalized effort: non-reasoning	#134 / 267	#38	$0.40	262K	Open weights	9 September 2026
54	Mercury 2.5Inception Labs	11.8%	Reasoning	#135 / 267	#174	$0.15	260K	No linked weights	23 September 2026
55	Gemma 4 26B A4BGoogle	11.5%	Non-reasoningSource label: Non-reasoningNormalized effort: non-reasoning	#137 / 267	#49	$0.225	262K	Open weights	9 September 2026
56	granite 4.2 30bIBM Granite	11.2%	Reasoning	#141 / 267	#175	—	131K	Open weights	5 September 2026
57	North Mini CodeCohere	11.1%	Reasoning	#142 / 267	#179	Free	256K	Open weights	5 September 2026
58	gpt-oss-20bOpenAI	11.0%	HighSource label: HighNormalized effort: high	#144 / 267	#205	$0.09	131K	Open weights	30 September 2026
59	Nanbeige4.1 3BNanbeige	10.9%	Reasoning	#146 / 267	—	—	—	Open weights	9 September 2026
60	Step3 VL 10BStepfun Ai	10.8%	Reasoning	#148 / 267	#66	—	—	Open weights	9 September 2026
61	Nemotron 3.5 LightningNVIDIA	10.6%	Reasoning	#152 / 267	#196	$0.20	262K	Restricted license	5 September 2026
62	Solar Open 100BUpstage	10.4%	ReasoningSource label: ReasoningNormalized effort: Unknown	#155 / 267	—	—	—	Restricted license	13 September 2026
63	Qwen3 Coder NextQwen	10.1%	Non-reasoning	#157 / 267	#77	$0.80	262K	Open weights	5 September 2026
64	Qwen3.5 4BQwen	9.9%	ReasoningSource label: ReasoningNormalized effort: Unknown	#159 / 267	#147	—	—	Open weights	13 September 2026
65	Granite 4.2 8BIBM Granite	9.7%	Reasoning	#161 / 267	#240	$0.25	131K	Open weights	5 September 2026
66	Ring flash 2.0InclusionAI	9.6%	Reasoning	#162 / 267	—	—	—	Open weights	12 September 2026
67	Ling 3.0 tinyInclusionAI	9.3%	Reasoning	#165 / 267	#137	—	—	Open weights	9 September 2026
68	MiniCPM5 2BOpenbmb	8.9%	Reasoning	#167 / 267	#145	—	—	Open weights	7 September 2026
69	Qwen3 Next 80B A3B InstructQwen	7.6%	Non-reasoning	#173 / 267	#104	$1.10	262K	Open weights	13 September 2026
70	granite 4.2 3bIBM Granite	6.6%	Reasoning	#182 / 267	#241	—	131K	Open weights	5 September 2026
71	MiniCPM5 1BOpenbmb	6.5%	ReasoningSource label: ReasoningNormalized effort: Unknown	#183 / 267	—	—	—	Open weights	13 September 2026
72	granite 4.0 h 350mIBM Granite	6.4%	Non-reasoning	#184 / 267	—	—	—	Open weights	12 September 2026
73	granite 4.0 350mIBM Granite	5.5%	Non-reasoning	#200 / 267	—	—	—	Open weights	12 September 2026
74	tiny aya globalCohereLabs	5.2%	Non-reasoning	#207 / 267	—	—	—	Gated access	5 September 2026
75	granite 4.0 h 1bIBM Granite	5.0%	Non-reasoning	#213 / 267	—	—	—	Open weights	12 September 2026
76	granite 4.0 microIBM Granite	5.0%	Non-reasoning	#213 / 267	—	—	—	Open weights	12 September 2026
77	Llama 4 MaverickMeta	4.9%	Non-reasoning	#220 / 267	#201	$0.652	1M	Gated access	5 September 2026
78	Qwen3 Omni 30B A3B InstructQwen	4.6%	Non-reasoning	#227 / 267	#117	—	—	Restricted license	13 September 2026
79	NVIDIA Nemotron Nano 9B v2NVIDIA	4.6%	Non-reasoningSource label: Non-reasoningNormalized effort: non-reasoning	#230 / 267	—	—	—	Restricted license	12 September 2026
80	Reka Flash 3Reka	4.4%	Reasoning	#237 / 267	—	$0.20	66K	Open weights	5 September 2026
81	Command ACohere	4.0%	Non-reasoning	#246 / 267	#155	$10.00	256K	Gated access	5 September 2026
82	granite 4.0 h smallIBM Granite	3.8%	Non-reasoning	#251 / 267	—	—	—	Open weights	12 September 2026
83	Llama 4 ScoutMeta	3.8%	Non-reasoning	#253 / 267	#206	$0.30	1.3M	Gated access	5 September 2026
84	Phi 4Microsoft	3.8%	Non-reasoning	#254 / 267	#289	$0.14	16K	Open weights	9 September 2026
85	Qwen3.5 2BQwen	2.6%	ReasoningSource label: ReasoningNormalized effort: Unknown	#265 / 267	—	—	—	Open weights	13 September 2026
86	Kimi Linear 48B A3B InstructMoonshot AI	2.5%	Non-reasoning	#266 / 267	—	—	—	Open weights	12 September 2026
87	Qwen3.5 0.8BQwen	1.1%	ReasoningSource label: ReasoningNormalized effort: Unknown	#267 / 267	#198	—	—	Open weights	13 September 2026

How ModelCap uses AA Humanity's Last Exam

AA Humanity's Last Exam ranks models by Humanity's Last Exam accuracy. ModelCap ingests the board as published, matches each entry to a catalogue model with a reviewed identity, and shows the source's score, interval and rank unchanged. AA Humanity's Last Exam is displayed as a public reference board and does not contribute to the ModelCap Index. Its published measurements and tested configurations remain visible here. The methodology page lists the sources admitted to the Index. Read the methodology and identity rules.

AA Humanity's Last Exam leaderboard: common questions

What is the AA Humanity's Last Exam benchmark?

Which AI model leads AA Humanity's Last Exam right now?

Claude Opus 5.5 (Anthropic) holds the top AA Humanity's Last Exam score among the models ModelCap tracks, at 61.4% as of the source snapshot published 16 October 2026.

How many models are ranked on the AA Humanity's Last Exam leaderboard here?

87 tracked models have a published AA Humanity's Last Exam result on ModelCap; the source board itself lists 267 entries. Each row shows the model's best evaluated configuration.

What is the best open-weight model on AA Humanity's Last Exam?

MiMo-V2.6-Pro from Xiaomi is the highest-scoring model with openly downloadable weights on this board, at 49.4%.

Which model offers the best value on AA Humanity's Last Exam?

Among the ten highest-scoring priced models, MiMo-V2.6-Pro has the lowest listed output price at $0.87 per 1M tokens while scoring 49.4%.

How is AA Humanity's Last Exam scored?

The board ranks models by Humanity's Last Exam accuracy; higher scores are better. ModelCap shows the source's own score, interval and rank and never re-runs the evaluation.

Does AA Humanity's Last Exam decide the ModelCap Index rank?

AA Humanity's Last Exam is displayed as a public reference board and does not contribute to the ModelCap Index. Its published measurements and tested configurations remain visible here. The methodology page lists the sources admitted to the Index.

How recent are the AA Humanity's Last Exam results?

The newest AA Humanity's Last Exam publication ModelCap holds is dated 16 October 2026. This page uses the published snapshot; source collection, publication and page updates happen separately.