Benchmark Almanac

SITE

GPT-4o (snapshot unspecified) 37.8% tops its leaderboard InternVL2.5-8B 32.8% leads the models we track
SenseNova-SI-1.3-InternVL3-8B 47.5% leads since January 2026
Fading

Models answer multiple-choice spatial questions over single images, multiple images and video, drawn from about 30 vision datasets. Tasks span figural to environmental scale.

Accepted at ICCV 2025. Some later papers report a chance-adjusted accuracy.

Released
May 2025
Kind
Spatial reasoning and counting
Level
Everyday
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Last in a launch post
January 2026, dropped before it saturated
Last model added
January 2026
Scores refreshed
November 12th, 2026

In the authors’ words

Spatial intelligence (SI) represents a cognitive ability encompassing the visualization, manipulation, and reasoning about spatial relationships, underpinning disciplines from neuroscience to robotics. We introduce SITE, a benchmark dataset towards SI Thorough Evaluation in a standardized format of multi-choice visual question-answering, designed to assess large vision-language models' spatial intelligence across diverse visual modalities (single-image, multi-image, and video) and SI factors (figural to environmental scales, spatial visualization and orientation, intrinsic and extrinsic, static and dynamic). Our approach to curating the benchmark combines a bottom-up survey about 31 existing datasets and a top-down strategy drawing upon three classification systems in cognitive science, which prompt us to design two novel types of tasks about view-taking and dynamic scenes. Extensive experiments reveal that leading models fall behind human experts especially in spatial orientation, a fundamental SI factor. Moreover, we demonstrate a positive correlation between a model's spatial reasoning proficiency and its performance on an embodied AI task.

Abstract of SITE: towards Spatial Intelligence Thorough Evaluation (arXiv:2505.05456, CC0)
9
Entries on its own leaderboard, 6 models we track
1
Launch post lists it, from 1 lab

GPT-4o (snapshot unspecified) tops the leaderboard with 37.8%

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is InternVL2.5-8B, at 32.8%.

SenseNova-SI-1.3-InternVL3-8B leads with 47.5%, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model.

The newest model with a score here came out in January 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 37.8 GPT-4o (snapshot unspecified)
  2. 32.8 InternVL2.5-8B
  3. 32.5 Gemini-1.5-Pro (snapshot unspecified)
  4. 31.4 Qwen2.5-VL-7B
  5. 30.2 LLaVA-OneVision 7B
  6. 29.5 Qwen2.5-VL-3B
  7. 29.4 InternVL2.5-4B
  8. 21.8 Phi-3.5-vision-instruct
  9. 18.4 LLAVA-OV-0.5B
  1. 1 GPT-4o (snapshot unspecified) 37.8%
  2. 2 InternVL2.5-8B 32.8%
  3. 3 Gemini-1.5-Pro (snapshot unspecified) 32.5%
  4. 4 Qwen2.5-VL-7B 31.4%
  5. 5 LLaVA-OneVision 7B 30.2%
  6. 6 Qwen2.5-VL-3B 29.5%
  7. 7 InternVL2.5-4B 29.4%
  8. 8 Phi-3.5-vision-instruct 21.8%
  9. 9 LLAVA-OV-0.5B 18.4%

Source: Official SITE leaderboard

Sources: Labs’ model cards and official SITE leaderboard

#EntryScoreSettingDateSource
1 GPT-4o (snapshot unspecified) 37.8% Official leaderboard
2 InternVL2.5-8B 32.8% Dec 2024 Official leaderboard
3 Gemini-1.5-Pro (snapshot unspecified) 32.5% Official leaderboard
4 Qwen2.5-VL-7B 31.4% Jan 2025 Official leaderboard
5 LLaVA-OneVision 7B 30.2% Aug 2024 Official leaderboard
6 Qwen2.5-VL-3B 29.5% Jan 2025 Official leaderboard
7 InternVL2.5-4B 29.4% Dec 2024 Official leaderboard
8 Phi-3.5-vision-instruct 21.8% Aug 2024 Official leaderboard
9 LLAVA-OV-0.5B 18.4% Official leaderboard
#ModelScoreSettingDateSource
1 SenseNova-SI-1.3-InternVL3-8B 47.5% Jan 2026 Lab model card
2 InternVL2.5-8B 32.8% Dec 2024 Official leaderboard
3 Qwen2.5-VL-7B 31.4% Jan 2025 Official leaderboard
4 LLaVA-OneVision 7B 30.2% Aug 2024 Official leaderboard
5 Qwen2.5-VL-3B 29.5% Jan 2025 Official leaderboard
6 InternVL2.5-4B 29.4% Dec 2024 Official leaderboard
7 Phi-3.5-vision-instruct 21.8% Aug 2024 Official leaderboard

Labs report scores for 1 model in their cards. Switch the scores above to rank them together.

Number one over time 2 models have held number one with claims included since its release in May 2025

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

Shanghai AI Lab
Own leaderboard Not tested since May 2025
InternVL2.5-8B 32.8%

Source: Official SITE leaderboard

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 Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) SenseNova-SI-1.3-InternVL3-8B InternVL2.5-8B LLaVA-OneVision 7B 0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) SenseNova-SI-1.3-InternVL3-8B InternVL2.5-8B

Sources: Official SITE leaderboard and labs’ model cards

Labs headline it less often now, last in January 2026

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

SenseTime
DateLaunch postScoreAs the post writes it
Jan 8, 2026 SenseNova-SI-1.3-InternVL3-8B 47.5 SITE

More in Vision

All 281 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. GPT-5.1 85.4%
  • MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
  • CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy. Qwen 3.8 Max 85.9%
  • ChartQA In use Models answer human-written and generated questions about chart images that need visual and arithmetic reasoning. Scored by relaxed accuracy with a 5 percent numeric tolerance. Command A Vision 90.9%