Humanity's Sixth Sense: Benchmarking Intuitive Visual Reasoning in Multimodal Models (Humanity's Sixth Sense (HSS))

We show this table for reference; we do not rank on it.

HSS tests what a model can infer from images and silent videos across temporal, spatial, social, and abstract reasoning. We show the published results as display-only evidence, excluded from overall and category rankings.

Pass@1 (%) on Humanity's Sixth Sense (HSS) — October 15, 2026 snapshot

We mirror the published pass@1 (%) view for Humanity's Sixth Sense (HSS). Gemini-4-Argon (high) leads the public snapshot at 55.4%, followed by GPT-6-Astra (max) (53.6%) and GPT-6.1-Sol (max) (46.6%). We do not use these results to rank models overall.

Gemini-4-Argon (high)

Google

55.4%

2Closed

GPT-6-Astra (max)

OpenAI

53.6%

Overall 83.88Context 1.05M

3Closed

GPT-6.1-Sol (max)

OpenAI

46.6%

Overall 84.07Context 1.05M

27 modelsMultimodal & GroundedCurrentDisplay onlyUpdated October 15, 2026 snapshot

Pass@1 (%) table (27 models)

Pass@1 (%) results for Humanity's Sixth Sense (HSS)
Rank	Model / configuration	Score	Parameters (B)	Open / closed
1	Gemini-4-Argon (high)Google	55.4% ±4.20 points	Not reported	Unknown
2	GPT-6-Astra (max)OpenAI	53.6% ±4.10 points	Not reported	Closed
3	GPT-6.1-Sol (max)OpenAI	46.6% ±3.80 points	Not reported	Closed
4	Claude-Opus-5.5 (xhigh)Anthropic	44.6% ±3.80 points	Not reported	Closed
5	Muse Wonder 1.0 (max)Meta	43.7% ±4.20 points	Not reported	Unknown
6	Gemini-3.8-Flash (high)Google	41.6% ±3.70 points	Not reported	Closed
7	Claude-Fable-5.1Anthropic	40.8% ±3.60 points	Not reported	Closed
8	Gemini-3.7-Flash (high)Google	39.8% ±3.70 points	Not reported	Closed
9	Muse Spark 1.3 (max)Meta	37.4% ±3.50 points	Not reported	Closed
10	Claude-Fable-5 (max)Anthropic	34.5% ±3.70 points	Not reported	Closed
11	Qwen3.8-Max (max)Alibaba	33% ±3.60 points	Not reported	Open
12	Gemini-3.5-Flash (high)Google	32.6% ±3.40 points	Not reported	Closed
13	Gemini-3.6-Flash (high)Google	31.9% ±3.40 points	Not reported	Closed
14	GPT-6-Sol (max)OpenAI	31.2% ±3.40 points	Not reported	Closed
15	Qwen3.8-Flash (max)Alibaba	30.9% ±3.30 points	Not reported	Open
16	Claude-Opus-5 (max)Anthropic	30.5% ±3.50 points	Not reported	Closed
17	GPT-5.6-Sol (max)OpenAI	30% ±3.30 points	Not reported	Closed
18	GLM-5.3-Flash (max)Z.AI	26.8% ±3.00 points	Not reported	Open
19	Kimi-K3 (max)Moonshot AI	25.5% ±3.10 points	Not reported	Pending
20	MiMo-v2.6-Flash (high)Xiaomi	25.4% ±2.90 points	Not reported	Open
21	MiMo-v2.6-Pro (high)Xiaomi	25.2% ±2.90 points	Not reported	Open
22	Claude-Sonnet-5 (max)Anthropic	24.8% ±3.20 points	Not reported	Closed
23	Muse Spark 1.2 (xhigh)Meta	24.5% ±3.10 points	Not reported	Closed
24	Claude-Opus-4.8 (max)Anthropic	23.5% ±3.20 points	Not reported	Closed
25	GPT-5.6-Terra (max)OpenAI	22.7% ±3.20 points	Not reported	Closed
26	GPT-5.6-Luna (max)OpenAI	21.6% ±3.20 points	Not reported	Closed
27	GPT-6-Luna (max)OpenAI	21% ±3.00 points	Not reported	Closed

How to read this leaderboard

Compare the published model configurations and their reasoning settings. The human baseline is 93.1%; overlapping uncertainty intervals do not establish a reliable ordering.

Operator receipt: 27 sourced rows are currently displayable on this page; the leading published row is Gemini-4-Argon (high) at 55.4%.

Honest limit: Video inputs exclude audio and transcripts. Claude Opus 5 grades the answers, and the scores depend on that judge. Public-web media may have appeared in training data; social judgments measure agreement with annotators. These results are display only and do not affect overall or category rankings.

HSS leaderboard HSS paper HSS dataset

How to read HSS results

We mirror 27 published rows from the HSS leaderboard, captured on October 15, 2026 snapshot, with each published model name and reasoning setting. The 522 tasks test inferences from images and silent videos.

Pass@1 averages three attempts per task. An answer passes only when it meets every rubric criterion. The published ± values are half-widths of 95% bootstrap intervals over tasks.

Video inputs exclude audio and transcripts. Claude Opus 5 grades the answers, and the scores depend on that judge. Public-web media may have appeared in training data; social judgments measure agreement with annotators. We keep the table display only and excluded from overall and category rankings.

Snapshot

27 published rows522 tasks288 image / 234 video4 domains / 11 subdomainsDisplay only

HSS leaderboard HSS paper HSS dataset

The published Humanity's Sixth Sense (HSS) snapshot places Gemini-4-Argon (high) first at 55.4%. The third row is 8.8 points behind. The broader top-10 range is 20.9 points, so the table still separates the published systems.

27 models have been evaluated on Humanity's Sixth Sense (HSS). The benchmark falls in the Multimodal & Grounded category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. Humanity's Sixth Sense (HSS) is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About Humanity's Sixth Sense (HSS)

Year

2026

Tasks

522 tasks (288 image, 234 video)

Format

Pass@1 (%) averaged over three attempts per task

Difficulty

Intuitive visual reasoning

The release contains 522 tasks: 288 images and 234 videos, organized into four domains and 11 subdomains. Pass@1 averages three attempts per task. An answer passes only when it meets every rubric criterion. The published ± values are half-widths of 95% bootstrap intervals over tasks.

Humanity's Sixth Sense paper Public benchmark source

Freshness and provenance

Version

Humanity's Sixth Sense (HSS) 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

Questions

What does Humanity's Sixth Sense (HSS) measure?

Which model leads the published Humanity's Sixth Sense (HSS) snapshot?

Gemini-4-Argon (high) currently leads the published Humanity's Sixth Sense (HSS) snapshot with 55.4% pass@1 (%). BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on Humanity's Sixth Sense (HSS)?

The October 15, 2026 snapshot snapshot contains 27 AI models.

Last updated: October 15, 2026 snapshot · mirrored from the public benchmark leaderboard

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 5,600+ readers.

One email each week. Unsubscribe anytime.