# Scientific Image Quality Assessment Understanding on SIQA-U Source: [Scientific Image Quality Assessment Understanding on SIQA-U]() Collection index: [llms.txt]() Data updated: 2026-03-10T08:00:30.612Z Evaluation of models on understanding scientific image quality, assessing metrics such as Scientific Completeness, Validity, Disciplinary Conformity, and Cognitive Clarity using the SIQA-U dataset. 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 Image Quality Assessment Understanding | | Dataset (including any stated split) | SIQA-U | | Metric | Scientific Completeness | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 14 of 14 | | 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 - [Scientific Completeness](): Higher is better; unit: PERCENT. - [Scientific Validity](): Higher is better; unit: PERCENT. - [Disciplinary Conformity](): Higher is better; unit: PERCENT. - [Cognitive Clarity](): Higher is better; unit: PERCENT. - [Yes/No Score](): Higher is better; unit: PERCENT. - [What Score](): Higher is better; unit: PERCENT. - [How Score](): Higher is better; unit: PERCENT. - [Overall Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | Scientific Completeness | Result date | Report paper | | --- | --- | --- | --- | | SIQA-Judger | 60.5 | Not recorded | 69afa57f2d603e6de3e43a08 | | InternVL3.5-241B | 47.9 | Not recorded | 69afa57f2d603e6de3e43a08 | | LLaMA-3.2-90B-Vision | 46.8 | Not recorded | 69afa57f2d603e6de3e43a08 | | Qwen3-VL-235B | 45.9 | Not recorded | 69afa57f2d603e6de3e43a08 | | GLM-4.6v | 45.8 | Not recorded | 69afa57f2d603e6de3e43a08 | | Gemini-2.5-Pro | 44.8 | Not recorded | 69afa57f2d603e6de3e43a08 | | GPT-5 | 44.1 | Not recorded | 69afa57f2d603e6de3e43a08 | | Doubao-1.5-Vision-Pro | 43.7 | Not recorded | 69afa57f2d603e6de3e43a08 | | Claude-sonnet-4.5 | 43 | Not recorded | 69afa57f2d603e6de3e43a08 | | GPT-4o | 42.7 | Not recorded | 69afa57f2d603e6de3e43a08 | | O3 | 42.4 | Not recorded | 69afa57f2d603e6de3e43a08 | | GPT-3.5-turbo | 39.9 | Not recorded | 69afa57f2d603e6de3e43a08 | | Random Guess | 36.5 | Not recorded | 69afa57f2d603e6de3e43a08 | | DeepSeek-VL2 | 33 | Not recorded | 69afa57f2d603e6de3e43a08 | ### SIQA-Judger Scientific Completeness: 60.5. Result ID: 69afce976dbb7075b957edc5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training":"trained"}. This record does not assert independent reproduction. ### InternVL3.5-241B Scientific Completeness: 47.9. Result ID: 69afce976dbb7075b957edc3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Open-Source"}. This record does not assert independent reproduction. ### LLaMA-3.2-90B-Vision Scientific Completeness: 46.8. Result ID: 69afce976dbb7075b957edc2. [Reporting paper]() Recorded evaluation configuration: {"Access":"Open-Source"}. This record does not assert independent reproduction. ### Qwen3-VL-235B Scientific Completeness: 45.9. Result ID: 69afce976dbb7075b957edc4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Open-Source"}. This record does not assert independent reproduction. ### GLM-4.6v Scientific Completeness: 45.8. Result ID: 69afce976dbb7075b957edc0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Open-Source"}. This record does not assert independent reproduction. ### Gemini-2.5-Pro Scientific Completeness: 44.8. Result ID: 69afce976dbb7075b957edbd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Closed-Source"}. This record does not assert independent reproduction. ### GPT-5 Scientific Completeness: 44.1. Result ID: 69afce976dbb7075b957edba. [Reporting paper]() Recorded evaluation configuration: {"Access":"Closed-Source"}. This record does not assert independent reproduction. ### Doubao-1.5-Vision-Pro Scientific Completeness: 43.7. Result ID: 69afce976dbb7075b957edbf. [Reporting paper]() Recorded evaluation configuration: {"Access":"Closed-Source"}. This record does not assert independent reproduction. ### Claude-sonnet-4.5 Scientific Completeness: 43. Result ID: 69afce976dbb7075b957edbe. [Reporting paper]() Recorded evaluation configuration: {"Access":"Closed-Source"}. This record does not assert independent reproduction. ### GPT-4o Scientific Completeness: 42.7. Result ID: 69afce976dbb7075b957edbb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Closed-Source"}. This record does not assert independent reproduction. ### O3 Scientific Completeness: 42.4. Result ID: 69afce976dbb7075b957edb9. [Reporting paper]() Recorded evaluation configuration: {"Access":"Closed-Source"}. This record does not assert independent reproduction. ### GPT-3.5-turbo Scientific Completeness: 39.9. Result ID: 69afce976dbb7075b957edbc. [Reporting paper]() Recorded evaluation configuration: {"Access":"Closed-Source"}. This record does not assert independent reproduction. ### Random Guess Scientific Completeness: 36.5. Result ID: 69afce976dbb7075b957edc6. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### DeepSeek-VL2 Scientific Completeness: 33. Result ID: 69afce976dbb7075b957edc1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Open-Source"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.