# Deepfake Detection on HydraFake Source: [Deepfake Detection on HydraFake]() Collection index: [llms.txt]() Data updated: 2026-03-24T15:42:27.458Z Evaluation of Deepfake Detection performance on the HydraFake dataset, reporting accuracy against various known forgery methods (e.g., ID, ADF, FLUX, GPT-4o) and an overall average. 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) | HydraFake | | Metric | ID Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 16 of 16 | | 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 - [ID Accuracy](): Higher is better; unit: PERCENT. - [ADF Accuracy](): Higher is better; unit: PERCENT. - [FLUX Accuracy](): Higher is better; unit: PERCENT. - [StarryAI Accuracy](): Higher is better; unit: PERCENT. - [MAGI-1 Accuracy](): Higher is better; unit: PERCENT. - [HART Accuracy](): Higher is better; unit: PERCENT. - [Infinity Accuracy](): Higher is better; unit: PERCENT. - [St.GAN2 Accuracy](): Higher is better; unit: PERCENT. - [ICLight Accuracy](): Higher is better; unit: PERCENT. - [CodeF. Accuracy](): Higher is better; unit: PERCENT. - [InfiniteY. Accuracy (Cross-Forgery)](): Higher is better; unit: PERCENT. - [PuLID Accuracy](): Higher is better; unit: PERCENT. - [FaceAda. Accuracy](): Higher is better; unit: PERCENT. - [Deepface. Accuracy](): Higher is better; unit: PERCENT. - [InfiniteY. Accuracy (Cross-Domain)](): Higher is better; unit: PERCENT. - [Dreamina Accuracy](): Higher is better; unit: PERCENT. - [HailuoAI Accuracy](): Higher is better; unit: PERCENT. - [GPT-4o Accuracy](): Higher is better; unit: PERCENT. - [FFIW Accuracy](): Higher is better; unit: PERCENT. - [Average Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | ID Accuracy | Result date | Report paper | | --- | --- | --- | --- | | VIGIL | 98.2 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Veritas | 97.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | DDA | 95.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | D^3 | 93.4 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Co-SPY | 88.9 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | C2P-CLIP | 87.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | NPR | 86.4 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | UnivFD | 83.2 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | FakeVLM | 78.5 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | AIDE | 77.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | SAFE | 73.4 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | GPT-5.2 | 71.9 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Gemini-3-Pro | 70 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Qwen3-VL-8B | 64.8 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | Kimi-K2.5 | 64.3 | Not recorded | 69c21a9f0636e5bf4afc49c3 | | GLM-4.6V | 60 | Not recorded | 69c21a9f0636e5bf4afc49c3 | ### VIGIL ID Accuracy: 98.2. Result ID: 69c2b0d147f34ce924abd743. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Veritas ID Accuracy: 97.3. Result ID: 69c2b0d147f34ce924abd742. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### DDA ID Accuracy: 95.3. Result ID: 69c2b0d147f34ce924abd73b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### D^3 ID Accuracy: 93.4. Result ID: 69c2b0d147f34ce924abd739. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Co-SPY ID Accuracy: 88.9. Result ID: 69c2b0d147f34ce924abd73a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### C2P-CLIP ID Accuracy: 87.3. Result ID: 69c2b0d147f34ce924abd738. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### NPR ID Accuracy: 86.4. Result ID: 69c2b0d147f34ce924abd735. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### UnivFD ID Accuracy: 83.2. Result ID: 69c2b0d147f34ce924abd734. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### FakeVLM ID Accuracy: 78.5. Result ID: 69c2b0d147f34ce924abd741. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### AIDE ID Accuracy: 77.3. Result ID: 69c2b0d147f34ce924abd737. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### SAFE ID Accuracy: 73.4. Result ID: 69c2b0d147f34ce924abd736. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GPT-5.2 ID Accuracy: 71.9. Result ID: 69c2b0d147f34ce924abd73f. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Gemini-3-Pro ID Accuracy: 70. Result ID: 69c2b0d147f34ce924abd740. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Qwen3-VL-8B ID Accuracy: 64.8. Result ID: 69c2b0d147f34ce924abd73c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Kimi-K2.5 ID Accuracy: 64.3. Result ID: 69c2b0d147f34ce924abd73e. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GLM-4.6V ID Accuracy: 60. Result ID: 69c2b0d147f34ce924abd73d. [Reporting 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.