SpatialViz-Bench
ReportedNo description yet: we only describe an eval set when we can do it correctly.
- Kind
- Published by a third party's leaderboard, collected by Epoch AI. The leaderboard is named and linked.
- Link
- https://arxiv.org/abs/2507.07610
- Score shown
- Epoch's column "Overall score", as a percentage, higher is better. The source gives no error bars.
- Results
- 8 in all: 3 for open-weight models, 4 for closed ones, 1 where Epoch does not say. 2 linked to a model page here.
Best open-weight result: meta-llama/Llama-4-Scout-17B-16E-Instruct, 34.2%, number 4 of 8 results.
| # | Model | Weights | Setting | Score | Date | Source |
|---|---|---|---|---|---|---|
| 1 | gemini-2.5-proGoogle DeepMind | closed | Prompting: CoT | 44.7% | not dated | arxiv.org |
| 2 | o1-2024-12-17_unknownOpenAI | closed | effort unknown; Prompting: CoT | 41.4% | not dated | arxiv.org |
| 3 | gemini-2.5-flashGoogle DeepMind | closed | Prompting: CoT | 36.9% | not dated | arxiv.org |
| 4 | meta-llama/Llama-4-Scout-17B-16E-InstructEpoch: Llama-4-Scout-17B-16E-Instruct, Meta AI | open | Prompting: CoT | 34.2% | not dated | arxiv.org |
| 5 | claude-3-7-sonnet-20250219_unknownAnthropic | closed | effort unknown; Prompting: CoT | 33.9% | not dated | arxiv.org |
| 6 | Qwen2.5-VL-72B-InstructAlibabano page match | open | Prompting: CoT | 33.3% | not dated | arxiv.org |
| 7 | Qwen-VL-Max | not stated | Prompting: CoT | 32.0% | not dated | arxiv.org |
| 8 | meta-llama/Llama-4-Maverick-17B-128E-InstructEpoch: Llama-4-Maverick-17B-128E-Instruct, Meta AI | open | Prompting: CoT | 31.8% | not dated | arxiv.org |
Open-weight rows are shaded. Dates are when the result was run or evaluated, where the data says. Data from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Epoch AI. Method