# Scientific Reasoning on Lens Source: [Scientific Reasoning on Lens]() Collection index: [llms.txt]() Data updated: 2026-05-14T08:25:59.159Z Evaluation of model performance on the Lens dataset for Scientific Reasoning tasks, measured using Accuracy. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Scientific Reasoning | | Dataset (including any stated split) | Lens | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 16 of 16 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | Gemini2.5-Pro | 56.2 | Not recorded | 68fa78b91e209131cea7a10f | | Qwen2.5-VL | 53.65 | Not recorded | 68fa78b91e209131cea7a10f | | Qwen2.5-VL | 51.66 | Not recorded | 68fa78b91e209131cea7a10f | | GPT-4o | 51.14 | Not recorded | 68fa78b91e209131cea7a10f | | QVQ-Max | 50.8 | Not recorded | 68fa78b91e209131cea7a10f | | InternVL3 | 49.39 | Not recorded | 68fa78b91e209131cea7a10f | | InternVL3 | 47.18 | Not recorded | 68fa78b91e209131cea7a10f | | Qwen2.5-VL | 46.28 | Not recorded | 68fa78b91e209131cea7a10f | | InternVL3 | 44.69 | Not recorded | 68fa78b91e209131cea7a10f | | Deepseek-VL2 | 44.58 | Not recorded | 68fa78b91e209131cea7a10f | | Gemma3 | 43.33 | Not recorded | 68fa78b91e209131cea7a10f | | InternVL3 | 40.56 | Not recorded | 68fa78b91e209131cea7a10f | | Qwen2.5-VL | 40.33 | Not recorded | 68fa78b91e209131cea7a10f | | Gemma3 | 39.53 | Not recorded | 68fa78b91e209131cea7a10f | | Deepseek-VL2-tiny | 38.97 | Not recorded | 68fa78b91e209131cea7a10f | | Kimi-VL-thinking | 29.4 | Not recorded | 68fa78b91e209131cea7a10f | ### Gemini2.5-Pro Accuracy: 56.2. Result ID: 6a05870447f34ce924ec7fd5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"-"}. This record does not assert independent reproduction. ### Qwen2.5-VL Accuracy: 53.65. Result ID: 6a05870447f34ce924ec7fe1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"72B"}. This record does not assert independent reproduction. ### Qwen2.5-VL Accuracy: 51.66. Result ID: 6a05870447f34ce924ec7fe0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"32B"}. This record does not assert independent reproduction. ### GPT-4o Accuracy: 51.14. Result ID: 6a05870447f34ce924ec7fd4. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model size":"-"}. This record does not assert independent reproduction. ### QVQ-Max Accuracy: 50.8. Result ID: 6a05870447f34ce924ec7fe2. [Reporting paper]() Recorded evaluation configuration: {"Model size":"72B"}. This record does not assert independent reproduction. ### InternVL3 Accuracy: 49.39. Result ID: 6a05870447f34ce924ec7fdd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"78B"}. This record does not assert independent reproduction. ### InternVL3 Accuracy: 47.18. Result ID: 6a05870447f34ce924ec7fdc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"38B"}. This record does not assert independent reproduction. ### Qwen2.5-VL Accuracy: 46.28. Result ID: 6a05870447f34ce924ec7fdf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"7B"}. This record does not assert independent reproduction. ### InternVL3 Accuracy: 44.69. Result ID: 6a05870447f34ce924ec7fdb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"9B"}. This record does not assert independent reproduction. ### Deepseek-VL2 Accuracy: 44.58. Result ID: 6a05870447f34ce924ec7fd7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"MoE 4.5B/27B"}. This record does not assert independent reproduction. ### Gemma3 Accuracy: 43.33. Result ID: 6a05870447f34ce924ec7fd9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"12B"}. This record does not assert independent reproduction. ### InternVL3 Accuracy: 40.56. Result ID: 6a05870447f34ce924ec7fda. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"2B"}. This record does not assert independent reproduction. ### Qwen2.5-VL Accuracy: 40.33. Result ID: 6a05870447f34ce924ec7fde. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"3B"}. This record does not assert independent reproduction. ### Gemma3 Accuracy: 39.53. Result ID: 6a05870447f34ce924ec7fd8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"4B"}. This record does not assert independent reproduction. ### Deepseek-VL2-tiny Accuracy: 38.97. Result ID: 6a05870447f34ce924ec7fd6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"MoE 1B/3B"}. This record does not assert independent reproduction. ### Kimi-VL-thinking Accuracy: 29.4. Result ID: 6a05870447f34ce924ec7fe3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model size":"MoE 2.8B/16B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.