# Deepfake Detection on OmniFake Cross-Architecture Source: [Deepfake Detection on OmniFake Cross-Architecture]() Collection index: [llms.txt]() Data updated: 2026-03-24T15:42:27.516Z Evaluation of Deepfake Detection models on the OmniFake Cross-Architecture dataset, reporting performance metrics across various source deepfake generation methods (FF++, FLUX, GPT-Image, Harmon). 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 | Deepfake Detection | | Dataset (including any stated split) | OmniFake Cross-Architecture | | Metric | Accuracy (FF++) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 17 of 17 | | 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 (FF++)](): Higher is better; unit: PERCENT. - [Accuracy (FLUX)](): Higher is better; unit: PERCENT. - [Accuracy (GPT-Image)](): Higher is better; unit: PERCENT. - [Accuracy (Harmon)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (FF++) | Result date | Report paper | | --- | --- | --- | --- | | DDA | 94 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | VIGIL | 92.8 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Veritas | 92.5 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | D^3 | 91.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | VIGIL Cold-Start | 90.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | UnivFD | 88 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | NPR | 84.9 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | AIDE | 82.4 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Co-SPY | 82.4 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Gemini-3-Pro | 81.5 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | C2P-CLIP | 80.8 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | FakeVLM | 79.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | GPT-5.2 | 77.4 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | SAFE | 74.9 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Kimi-K2.5 | 66.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | GLM-4.6V | 54.1 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Qwen3-VL-8B | 50.7 | Not recorded | 69c21a9f0636e5bf4afc49c3 | ### DDA Accuracy (FF++): 94. Result ID: 69c2b0d147f34ce924abd75c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### VIGIL Accuracy (FF++): 92.8. Result ID: 69c2b0d147f34ce924abd765. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B"}. This record does not assert independent reproduction. ### Veritas Accuracy (FF++): 92.5. Result ID: 69c2b0d147f34ce924abd763. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### D^3 Accuracy (FF++): 91.3. Result ID: 69c2b0d147f34ce924abd75a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### VIGIL Cold-Start Accuracy (FF++): 90.3. Result ID: 69c2b0d147f34ce924abd764. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","mode":"Cold-Start"}. This record does not assert independent reproduction. ### UnivFD Accuracy (FF++): 88. Result ID: 69c2b0d147f34ce924abd755. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### NPR Accuracy (FF++): 84.9. Result ID: 69c2b0d147f34ce924abd756. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### AIDE Accuracy (FF++): 82.4. Result ID: 69c2b0d147f34ce924abd758. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Co-SPY Accuracy (FF++): 82.4. Result ID: 69c2b0d147f34ce924abd75b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Gemini-3-Pro Accuracy (FF++): 81.5. Result ID: 69c2b0d147f34ce924abd761. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### C2P-CLIP Accuracy (FF++): 80.8. Result ID: 69c2b0d147f34ce924abd759. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### FakeVLM Accuracy (FF++): 79.3. Result ID: 69c2b0d147f34ce924abd762. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GPT-5.2 Accuracy (FF++): 77.4. Result ID: 69c2b0d147f34ce924abd760. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### SAFE Accuracy (FF++): 74.9. Result ID: 69c2b0d147f34ce924abd757. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Kimi-K2.5 Accuracy (FF++): 66.3. Result ID: 69c2b0d147f34ce924abd75f. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GLM-4.6V Accuracy (FF++): 54.1. Result ID: 69c2b0d147f34ce924abd75e. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Qwen3-VL-8B Accuracy (FF++): 50.7. Result ID: 69c2b0d147f34ce924abd75d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"zero-shot":"true"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.