# Metric Depth Estimation on DepthVLM-Bench 1.0 (test) Source: [Metric Depth Estimation on DepthVLM-Bench 1.0 (test)]() Collection index: [llms.txt]() Data updated: 2026-05-18T08:28:12.416Z Evaluation of Metric Depth Estimation performance across multiple datasets (including Argoverse2, Waymo, NYUv2, etc.) using the DepthVLM-Bench 1.0 protocol on the test split, reporting delta1 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 | Metric Depth Estimation | | Dataset (including any stated split) | DepthVLM-Bench 1.0 (test) | | Metric | Delta1 Accuracy (Argoverse2) | | 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 - [Delta1 Accuracy (Argoverse2)](): Higher is better; unit: PERCENT. - [Delta1 Accuracy (Waymo)](): Higher is better; unit: PERCENT. - [Delta1 Accuracy (DDAD)](): Higher is better; unit: PERCENT. - [Delta1 Accuracy (NuScenes)](): Higher is better; unit: PERCENT. - [Delta1 Accuracy (ETH3D)](): Higher is better; unit: PERCENT. - [Delta1 Accuracy (ScanNet++)](): Higher is better; unit: PERCENT. - [Delta1 Accuracy (sunRGBD)](): Higher is better; unit: PERCENT. - [Delta1 Accuracy (IBims-1)](): Higher is better; unit: PERCENT. - [Delta1 Accuracy (NYUv2)](): Higher is better; unit: PERCENT. - [Delta1 Accuracy (Average)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Delta1 Accuracy (Argoverse2) | Result date | Report paper | | --- | --- | --- | --- | | Ours-4B | 81 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | Ours-8B | 79.8 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | DepthLM-12B | 76.1 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | Youtu-VL-4B | 66.3 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | GPT-5.5 | 37.8 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | GPT-4o | 14.1 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | InternVL3.5-8B | 13.9 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | InternVL3.5-38B | 13.1 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | Qwen3-VL-8B | 11.9 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | InternVL3.5-14B | 8.6 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | SpatialRGPT-8B | 4.5 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | Qwen3-VL-4B | 3.3 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | Qwen3-VL-32B | 2.9 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | SpaceLLaVA-13B | 0.6 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | Cambrian-S-7B | 0.6 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | | Always Output 2.0m | 0.2 | Not recorded | 6a0a9cecdbc836b87d82b2f8 | ### Ours-4B Delta1 Accuracy (Argoverse2): 81. Result ID: 6a0acd3c47f34ce924f0e3c5. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Category":"VLMs Trained on Metric Depth Estimation","Backbone":"Qwen3-VL-4B"}. This record does not assert independent reproduction. ### Ours-8B Delta1 Accuracy (Argoverse2): 79.8. Result ID: 6a0acd3c47f34ce924f0e3c6. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Category":"VLMs Trained on Metric Depth Estimation","Backbone":"Qwen3-VL-8B"}. This record does not assert independent reproduction. ### DepthLM-12B Delta1 Accuracy (Argoverse2): 76.1. Result ID: 6a0acd3c47f34ce924f0e3c4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"VLMs Trained on Metric Depth Estimation"}. This record does not assert independent reproduction. ### Youtu-VL-4B Delta1 Accuracy (Argoverse2): 66.3. Result ID: 6a0acd3c47f34ce924f0e3c3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Category":"VLMs Trained on Metric Depth Estimation"}. This record does not assert independent reproduction. ### GPT-5.5 Delta1 Accuracy (Argoverse2): 37.8. Result ID: 6a0acd3c47f34ce924f0e3b9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"General-Purpose VLMs","Prompting Strategy":"DepthLM \[9\]"}. This record does not assert independent reproduction. ### GPT-4o Delta1 Accuracy (Argoverse2): 14.1. Result ID: 6a0acd3c47f34ce924f0e3b8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"General-Purpose VLMs","Prompting Strategy":"DepthLM \[9\]"}. This record does not assert independent reproduction. ### InternVL3.5-8B Delta1 Accuracy (Argoverse2): 13.9. Result ID: 6a0acd3c47f34ce924f0e3bd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"General-Purpose VLMs","Prompting Strategy":"DepthLM \[9\]"}. This record does not assert independent reproduction. ### InternVL3.5-38B Delta1 Accuracy (Argoverse2): 13.1. Result ID: 6a0acd3c47f34ce924f0e3bf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"General-Purpose VLMs","Prompting Strategy":"DepthLM \[9\]"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Delta1 Accuracy (Argoverse2): 11.9. Result ID: 6a0acd3c47f34ce924f0e3bb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"General-Purpose VLMs","Prompting Strategy":"DepthLM \[9\]"}. This record does not assert independent reproduction. ### InternVL3.5-14B Delta1 Accuracy (Argoverse2): 8.6. Result ID: 6a0acd3c47f34ce924f0e3be. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"General-Purpose VLMs","Prompting Strategy":"DepthLM \[9\]"}. This record does not assert independent reproduction. ### SpatialRGPT-8B Delta1 Accuracy (Argoverse2): 4.5. Result ID: 6a0acd3c47f34ce924f0e3c1. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Category":"Spatial-Enhanced VLMs","Prompting Strategy":"DepthLM \[9\]"}. This record does not assert independent reproduction. ### Qwen3-VL-4B Delta1 Accuracy (Argoverse2): 3.3. Result ID: 6a0acd3c47f34ce924f0e3ba. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"General-Purpose VLMs","Prompting Strategy":"DepthLM \[9\]"}. This record does not assert independent reproduction. ### Qwen3-VL-32B Delta1 Accuracy (Argoverse2): 2.9. Result ID: 6a0acd3c47f34ce924f0e3bc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"General-Purpose VLMs","Prompting Strategy":"DepthLM \[9\]"}. This record does not assert independent reproduction. ### SpaceLLaVA-13B Delta1 Accuracy (Argoverse2): 0.6. Result ID: 6a0acd3c47f34ce924f0e3c0. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Category":"Spatial-Enhanced VLMs","Prompting Strategy":"DepthLM \[9\]"}. This record does not assert independent reproduction. ### Cambrian-S-7B Delta1 Accuracy (Argoverse2): 0.6. Result ID: 6a0acd3c47f34ce924f0e3c2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"Spatial-Enhanced VLMs","Prompting Strategy":"DepthLM \[9\]"}. This record does not assert independent reproduction. ### Always Output 2.0m Delta1 Accuracy (Argoverse2): 0.2. Result ID: 6a0acd3c47f34ce924f0e3b7. [Reporting paper]() Recorded evaluation configuration: {"Category":"Naive Prediction with Constant Answers"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.