# Multimodal Deep Search on MTA (test) Source: [Multimodal Deep Search on MTA (test)]() Collection index: [llms.txt]() Data updated: 2026-04-09T23:17:58.418Z Evaluation of model performance on the MTA (test) dataset for Multimodal Deep Search tasks, measured by 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 | Multimodal Deep Search | | Dataset (including any stated split) | MTA (test) | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 12 of 12 | | 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 | | --- | --- | --- | --- | | MTA-DeepSearch-32B | 29.78 | Not recorded | 69d732707d0172140ec76a0c | | Gemini-3 Pro | 28.65 | Not recorded | 69d732707d0172140ec76a0c | | Gemini-2.5 Pro | 27.53 | Not recorded | 69d732707d0172140ec76a0c | | Gemini-3-pro | 26.97 | Not recorded | 69d732707d0172140ec76a0c | | GPT-5 | 25.84 | Not recorded | 69d732707d0172140ec76a0c | | Gemini-2.5-pro | 21.91 | Not recorded | 69d732707d0172140ec76a0c | | GPT-5 | 21.35 | Not recorded | 69d732707d0172140ec76a0c | | MTA-DeepSearch-8B | 20.79 | Not recorded | 69d732707d0172140ec76a0c | | Qwen3-VL-32B-Inst. | 17.42 | Not recorded | 69d732707d0172140ec76a0c | | Qwen3-VL-8B-Inst. | 11.8 | Not recorded | 69d732707d0172140ec76a0c | | Qwen3-VL-32B-Inst. | 11.24 | Not recorded | 69d732707d0172140ec76a0c | | Qwen3-VL-8B-Inst. | 6.18 | Not recorded | 69d732707d0172140ec76a0c | ### MTA-DeepSearch-32B Accuracy: 29.78. Result ID: 69d8334c47f34ce924bfdf19. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Ours","Model Scale":"32B"}. This record does not assert independent reproduction. ### Gemini-3 Pro Accuracy: 28.65. Result ID: 69d8334c47f34ce924bfdf15. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Agent Workflow"}. This record does not assert independent reproduction. ### Gemini-2.5 Pro Accuracy: 27.53. Result ID: 69d8334c47f34ce924bfdf14. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Agent Workflow"}. This record does not assert independent reproduction. ### Gemini-3-pro Accuracy: 26.97. Result ID: 69d8334c47f34ce924bfdf10. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Direct Answer"}. This record does not assert independent reproduction. ### GPT-5 Accuracy: 25.84. Result ID: 69d8334c47f34ce924bfdf13. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Agent Workflow"}. This record does not assert independent reproduction. ### Gemini-2.5-pro Accuracy: 21.91. Result ID: 69d8334c47f34ce924bfdf0f. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Direct Answer"}. This record does not assert independent reproduction. ### GPT-5 Accuracy: 21.35. Result ID: 69d8334c47f34ce924bfdf0e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Direct Answer"}. This record does not assert independent reproduction. ### MTA-DeepSearch-8B Accuracy: 20.79. Result ID: 69d8334c47f34ce924bfdf18. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Ours","Model Scale":"8B"}. This record does not assert independent reproduction. ### Qwen3-VL-32B-Inst. Accuracy: 17.42. Result ID: 69d8334c47f34ce924bfdf17. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Agent Workflow","Model Scale":"32B"}. This record does not assert independent reproduction. ### Qwen3-VL-8B-Inst. Accuracy: 11.8. Result ID: 69d8334c47f34ce924bfdf16. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Agent Workflow","Model Scale":"8B"}. This record does not assert independent reproduction. ### Qwen3-VL-32B-Inst. Accuracy: 11.24. Result ID: 69d8334c47f34ce924bfdf12. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Direct Answer","Model Scale":"32B"}. This record does not assert independent reproduction. ### Qwen3-VL-8B-Inst. Accuracy: 6.18. Result ID: 69d8334c47f34ce924bfdf11. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Direct Answer","Model Scale":"8B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.