# Deepfake Detection on OmniFake Cross-Model Source: [Deepfake Detection on OmniFake Cross-Model]() Collection index: [llms.txt]() Data updated: 2026-03-24T15:42:27.551Z Evaluation of Deepfake Detection performance on the OmniFake Cross-Model split, reporting accuracy metrics specific to generative models including Midjourney, SD3, Sora2, and Veo3. 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-Model | | Metric | Midjourney Accuracy | | 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 - [Midjourney Accuracy](): Higher is better; unit: PERCENT. - [NOVA Accuracy](): Higher is better; unit: PERCENT. - [Nano Accuracy](): Higher is better; unit: PERCENT. - [SD3 Accuracy](): Higher is better; unit: PERCENT. - [Sora2 Accuracy](): Higher is better; unit: PERCENT. - [Veo3 Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Midjourney Accuracy | Result date | Report paper | | --- | --- | --- | --- | | VIGIL | 99.4 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | DDA | 97.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | VIGIL Cold-Start | 96.2 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Co-SPY | 94.5 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Gemini-3-Pro | 91.6 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | AIDE | 91.1 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | GPT-5.2 | 90.8 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Veritas | 89.5 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Kimi-K2.5 | 87.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | D^3 | 86.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Qwen3-VL-8B | 84.9 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | FakeVLM | 77.5 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | GLM-4.6V | 72.1 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | UnivFD | 66.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | SAFE | 65.2 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | C2P-CLIP | 63.4 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | NPR | 62.9 | Not recorded | 69c21a9f0636e5bf4afc49c3 | ### VIGIL Midjourney Accuracy: 99.4. Result ID: 69c2b0d147f34ce924abd776. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B"}. This record does not assert independent reproduction. ### DDA Midjourney Accuracy: 97.3. Result ID: 69c2b0d147f34ce924abd76d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### VIGIL Cold-Start Midjourney Accuracy: 96.2. Result ID: 69c2b0d147f34ce924abd775. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","mode":"Cold-Start"}. This record does not assert independent reproduction. ### Co-SPY Midjourney Accuracy: 94.5. Result ID: 69c2b0d147f34ce924abd76c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Gemini-3-Pro Midjourney Accuracy: 91.6. Result ID: 69c2b0d147f34ce924abd772. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### AIDE Midjourney Accuracy: 91.1. Result ID: 69c2b0d147f34ce924abd769. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GPT-5.2 Midjourney Accuracy: 90.8. Result ID: 69c2b0d147f34ce924abd771. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Veritas Midjourney Accuracy: 89.5. Result ID: 69c2b0d147f34ce924abd774. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Kimi-K2.5 Midjourney Accuracy: 87.3. Result ID: 69c2b0d147f34ce924abd770. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### D^3 Midjourney Accuracy: 86.3. Result ID: 69c2b0d147f34ce924abd76b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Qwen3-VL-8B Midjourney Accuracy: 84.9. Result ID: 69c2b0d147f34ce924abd76e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"zero-shot":"true"}. This record does not assert independent reproduction. ### FakeVLM Midjourney Accuracy: 77.5. Result ID: 69c2b0d147f34ce924abd773. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GLM-4.6V Midjourney Accuracy: 72.1. Result ID: 69c2b0d147f34ce924abd76f. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### UnivFD Midjourney Accuracy: 66.3. Result ID: 69c2b0d147f34ce924abd766. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### SAFE Midjourney Accuracy: 65.2. Result ID: 69c2b0d147f34ce924abd768. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### C2P-CLIP Midjourney Accuracy: 63.4. Result ID: 69c2b0d147f34ce924abd76a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### NPR Midjourney Accuracy: 62.9. Result ID: 69c2b0d147f34ce924abd767. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.