# Multimodal Retrieval on M-BEIR extended (test) Source: [Multimodal Retrieval on M-BEIR extended (test)]() Collection index: [llms.txt]() Data updated: 2026-03-04T16:57:01.305Z Evaluation of Multimodal Retrieval performance using the M-BEIR extended test split, reporting cross-modal recall scores across various modality pairs (Image, Text, Video, Audio). 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 Retrieval | | Dataset (including any stated split) | M-BEIR extended (test) | | Metric | Recall (I→I) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 9 of 9 | | 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 - [Recall (I→I)](): Higher is better; unit: PERCENT. - [Recall (T→T)](): Higher is better; unit: PERCENT. - [Recall (I→T)](): Higher is better; unit: PERCENT. - [Recall (T→I)](): Higher is better; unit: PERCENT. - [Recall (V→T)](): Higher is better; unit: PERCENT. - [Recall (T→V)](): Higher is better; unit: PERCENT. - [Recall (A→T)](): Higher is better; unit: PERCENT. - [Recall (T→A)](): Higher is better; unit: PERCENT. - [Recall (T→I,T)](): Higher is better; unit: PERCENT. - [Recall (I,T→T)](): Higher is better; unit: PERCENT. - [Recall (I,T→I)](): Higher is better; unit: PERCENT. - [Recall (I,T→I,T)](): Higher is better; unit: PERCENT. - [Recall (V,T→V)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Recall (I→I) | Result date | Report paper | | --- | --- | --- | --- | | MMEmbed | 32.1 | Not recorded | 69a66c572d603e6de3e3d146 | | PE-Core | 32 | Not recorded | 69a66c572d603e6de3e3d146 | | VLM2VecV2 | 30 | Not recorded | 69a66c572d603e6de3e3d146 | | CLIP\_SF | 28.4 | Not recorded | 69a66c572d603e6de3e3d146 | | CLIP | 25.9 | Not recorded | 69a66c572d603e6de3e3d146 | | SigLIP | 25.9 | Not recorded | 69a66c572d603e6de3e3d146 | | OmniRet | 24.4 | Not recorded | 69a66c572d603e6de3e3d146 | ### MMEmbed Recall (I→I): 32.1. Result ID: 69a8644e6dbb7075b9500ec9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"7B"}. This record does not assert independent reproduction. ### PE-Core Recall (I→I): 32. Result ID: 69a8644e6dbb7075b9500ec5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"-"}. This record does not assert independent reproduction. ### VLM2VecV2 Recall (I→I): 30. Result ID: 69a8644e6dbb7075b9500eca. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"1.5B"}. This record does not assert independent reproduction. ### CLIP\_SF Recall (I→I): 28.4. Result ID: 69a8644e6dbb7075b9500ec8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"-"}. This record does not assert independent reproduction. ### CLIP Recall (I→I): 25.9. Result ID: 69a8644e6dbb7075b9500ec3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM Size":"-"}. This record does not assert independent reproduction. ### SigLIP Recall (I→I): 25.9. Result ID: 69a8644e6dbb7075b9500ec4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"-"}. This record does not assert independent reproduction. ### OmniRet Recall (I→I): 24.4. Result ID: 69a8644e6dbb7075b9500ecb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"1.5B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.