VGGT-1B-Depth is a hosted model on Replicate that estimates scene depth from still images or video. Upload related views or a clip, run inference, then inspect the resulting depth images, per-image JSON, and—if enabled—a GLB point-cloud visualization. It is a way to explore 3D scene geometry, not a guarantee of a metrically accurate scan or finished reconstruction.
What VGGT-1B-Depth is
VGGT stands for Visual Geometry Grounded Transformer. The upstream project describes VGGT as a feed-forward neural network that predicts scene geometry from one, a few, or many views. The broader model family can infer camera parameters, depth maps, point maps, and 3D point tracks in a single forward pass. See the upstream VGGT repository and the CVPR 2025 paper.
As an Amazon Associate I earn from qualifying purchases.
The Replicate listing, vufinder/vggt-1b-depth, is a particular depth-focused implementation, not a promise that every VGGT deployment returns the same set of outputs. Its page says it uses VGGT-1B-Commercial weights and omits the point-map head because the depth head makes that head redundant in this variant. The upstream project describes the broader family as predicting camera parameters, depth maps, point maps, and 3D point tracks; the hosted depth variant exposes a narrower output set. Check the live Replicate model listing for its current behavior.
Recommended Free Tools
What you can upload
The Replicate interface lists JPG, JPEG, PNG, and WEBP images, along with MP4, AVI, and MOV video. You can submit multiple files, which makes related views of a scene a practical starting point. For video, a sampling-rate setting controls how often frames are selected; the listing says the first and last frames are always included.
#1 Best Overall
- System Compatibility Note: This 2-slot card measures 271 x 112 x 39 mm and requires a single 12V-2x6-pin power connector. Please verify chassis and PSU compatibility before purchase.
- Dedicated Support: Please contact us directly through Amazon for any product questions or assistance you may require.
- Professional Intel Arc Pro B70 GPU: Built on the Intel Xe2-HPG architecture, it features 32 Xe cores and 256 XMX engines, designed to accelerate AI, rendering, and complex visualization workloads.
- Massive 32GB GDDR6 VRAM: Equipped with 32GB of high-speed GDDR6 memory on a 256-bit bus, running at 19 Gbps, which allows for handling large AI models and complex datasets locally.
- High-Performance Engine Clock: Delivers an engine clock of 2540 MHz, providing the compute power needed for demanding professional applications and AI inference.
- Still images: use a small set of views that show the same scene from different positions.
- Video: choose a sampling rate appropriate to the clip; more sampled frames mean more input to process.
The listing says that input images and sampled video frames are padded to one aspect ratio and resized so their maximum dimension is 518 pixels. That is documented preprocessing, not a resolution control you can tune. Other listed settings include whether to return point-cloud and depth-image outputs, options for base64-encoded JSON arrays and alpha-channel blending, and the ability to exclude selected output keys. Consult the current listing for the exact controls and accepted values.
How to run it on Replicate
- Open the model page: go to vufinder/vggt-1b-depth on Replicate and review its current inputs and output options.
- Choose a test input: upload a few related scene images, or a short video in a supported format. For video, set the frame-sampling rate.
- Select outputs: leave depth images enabled if you want visual distance maps; enable the point cloud if you want a GLB visualization. The page also provides JSON data.
- Run the model: submit the input through Replicate’s web interface and wait for the prediction to finish.
- Inspect the results: open the depth images and JSON, and download the GLB if you want to view the scene in compatible 3D software.
For scripts, the listing documents examples for Python, Node.js, and HTTP, along with a pinned model version. Programmatic calls require a Replicate API token; keep it private and do not place it in public code or a shared notebook. Because the version identifier and hosted pricing can change, take the current values and instructions from the live model page rather than relying on a copied example.
Rank #2
- PLEASE NOTE: Exporting an NVIDIA RTX Pro 6000 GPU outside the US requires strict adherence to the U.S. Export Administration Regulations (EAR) and issuance of an export license from the Bureau of Industry and Security (BIS). Compliance and Know Your Customer (KYC) screening may be required as a condition of order acceptance. [NVIDIA Blackwell Streaming Multiprocessor] The new SM features increased processing throughput, and new neural shaders that integrate neural networks inside of programmable shaders | DLSS 4: Multi Frame Generation ensures ultra-smooth frame pacing for lifelike simulations.
- [Double-Flow-Through Design] The RTX PRO 6000 Blackwell features a double-flow-through cooling design, optimizing efficiency and airflow to sustain peak performance under 600W power loads. | [5th Gen Tensor Cores] Deliver up to 3X the performance of the previous generation and support for FP4 precision for faster AI model processing times with reduced memory usage, enabling local fine-tuning of LLMs and generative AI | [4th Gen Ray Tracing Cores] Double the ray-triangle intersection rate of the previous generation to create photoreal, physically accurate scenes and immersive 3D designs with RTX Mega Geometry, which enables up to 100X more ray-traced triangles.
- [PCIe Gen 5] Support for PCIe Gen 5 provides double the bandwidth of PCIe Gen 4, improving data-transfer speeds from CPU memory and unlocking faster performance for data-intensive tasks like AI, data science, and 3D modeling. | [GDDR7 Memory] With 96 GB of GPU memory and 1.8 TB ps bandwidth, it can tackle massive 3D and AI projects, fine-tune AI models locally, explore large-scale VR environments, and drive larger multi-app workflows.
- [DisplayPort 2.1] Achieve unparalleled visual clarity and performance, driving high resolution displays at up to 8K at 240 Hz and 16K at 60 Hz. Increased bandwidth enables seamless multi-monitor setups while HDR and higher color depth support ensures superior color accuracy for precision work, such as video editing, 3D design, and live broadcasting.
- [Universal MIG] Divide a single RTX PRO 6000 Blackwell into multiple isolated instances, each with dedicated resources, allowing for concurrent execution of multiple workloads, optimized GPU utilization, and secure isolation of different applications or users. [WARRANTY] 3 YR Manufacturer's Warranty. Bulk OEM Packaging. Retail Packaging is NOT included.
How to understand the outputs
| Output | What it contains | How to inspect it |
|---|---|---|
| Depth images | Visual representations of estimated distance structure. | Open them as images to see how the model estimates depth across the scene. Treat them as model outputs, not a verified measurement. |
| Per-image JSON | Numerical data for each image, including the image, pose encoding, depth, depth confidence, original image dimensions, and an optional mask. | Use a JSON viewer or code to inspect or process the per-image values. The listing spells one dimension field as heigth; that appears to be a schema/documentation typo for height, not a different dimension. |
| Optional GLB point cloud | A GLB file containing a point cloud and camera meshes. | Open it in a compatible 3D viewer for a convenient spatial visualization. |
These outputs help you inspect estimated geometry, but the cited project and model pages do not establish that the result is a production-ready reconstruction or a guaranteed metrically accurate scan.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Hosted inference or a local setup?
Replicate is the simpler route if you want to try the model without setting up its Python environment. The upstream repository documents local inference using Python, PyTorch, torchvision, NumPy, Pillow, and Hugging Face Hub dependencies, with examples for loading the pretrained model and preparing images. That is a developer workflow rather than a one-click desktop installation. The available documentation does not establish a dependable minimum GPU-memory requirement, so a specific local computer cannot be recommended from it.
Rank #3
- System Compatibility Note: 2-slot card, 271x112x39mm, single 8-pin power, 200W TDP. Verify chassis clearance and PSU capacity before purchase.
- Dedicated Support: Please contact us directly through Amazon for any product questions or assistance you may require.
- 24GB GDDR6 on 192-Bit Bus: Massive 24GB memory with 456 GB/s bandwidth – ideal for LLMs, AI inference, 3D rendering, and generative design.
- Intel Xe2-HPG Architecture: Built on Intel's next-gen architecture with 20 Xe cores and 160 XMX engines for AI acceleration (197 INT8 TOPS).
- PCIe 5.0 Support: PCI Express 5.0 x16 interface for maximum bandwidth with the latest workstation platforms.
| Consideration | Replicate-hosted run | Upstream local workflow |
|---|---|---|
| Setup | Upload inputs and configure the hosted model’s controls in Replicate. | Install and manage the documented Python dependencies, load the model, and prepare inputs. |
| Runtime control | Uses the hosted version and settings offered on the model listing. | Gives a developer more control over the local runtime and code. |
| Outputs | The listing exposes depth images, JSON, and an optional GLB point-cloud file. | Use the repository’s inference examples to work with model predictions; output details depend on the code path used. |
| Cost and hardware | Replicate displays an approximate, input-dependent run cost on its live listing. | Requires suitable local compute; the cited documentation does not specify a minimum GPU-memory figure. |
| Checkpoint and terms | The Replicate page identifies its weights as VGGT-1B-Commercial. | Confirm which checkpoint you load and the license terms that apply. |
The CVPR paper describes prediction from one, a few, or hundreds of views and says inference is performed “in seconds.” That is a general description of the model in the paper, not a runtime guarantee for every input or for Replicate’s hosted deployment.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Checkpoint licensing and commercial use
The Replicate page identifies the hosted weights as VGGT-1B-Commercial. The upstream project’s README, updated July 29, 2025, says that this commercial checkpoint permits commercial use subject to its license, while the original checkpoint remains non-commercial. Before a commercial deployment or redistribution, verify the exact checkpoint and read the current VGGT license; this guide is not a substitute for interpreting its terms.
Rank #4
- 【High-Performance APU】The MS-S1 MAX features an AMD Ryzen AI Max+ 395 APU, integrating a Zen 5 architecture CPU (up to 5.1GHz, 16C/32T, 64M L3 Cache), an RDNA 3.5 GPU, and an NPU (50 TOPS). The total system output is 126 TOPS. It provides powerful parallel computing capabilities for demanding AI workflows. It is ideal for running local LLMs, multimodal models, and computationally intensive tasks
- 【128GB UMA Memory】Equipped with up to 128GB of LPDDR5x-8000MT/s unified memory, it enables the CPU and GPU to access a shared, high-bandwidth memory pool with extremely low latency. Ideal for large-scale AI inference, 3D workloads, and complex timelines in video editing. It eliminates traditional VRAM bottlenecks, ensuring smoother data transfer during high-intensity computations. The UMA design maximizes performance stability under high loads
- 【Flexible Expansion】The MS-S1 MAX features USB4 V2 (up to 80Gbps), dual 10GbE LAN, HDMI 2.1 (up to 8K60), a full-length PCIe x16 expansion slot, and dual M.2 slots supporting up to 16TB RAID 0/1. Wi-Fi 7 provides stronger signal coverage and a more stable wireless experience. The slide-out design facilitates upgrades and maintenance. It easily adapts to personal, studio, or rack-mount enterprise environments
- 【High-Efficiency Cooling System】Utilizing an aerospace-grade aluminum alloy chassis, copper base plate, six heat pipes, dual turbine fans, and advanced PCM thermal conductive material, it maintains stable cooling performance even under continuous load. This system supports 130W continuous power and 160W peak power operation, with a built-in 320W power supply. It boasts multiple global certifications including CCC, FCC, UL, CE, and UKCA, ensuring stable and reliable operation in various environments
- 【Cluster Design】Two MS-S1 MAX units can be configured as a dual-unit cluster to run a large 235B Q4 model locally, achieving an output speed of 10.87 tok/s. Supporting 2U rack deployment, multiple MS-S1 MAX units can be cascaded into a distributed cluster to create a high-efficiency AI computing center. A cluster of four MS-S1 MAX units successfully ran a DeepSeek-R1 671B Q4 large model. A reserved cluster power-on interface allows for unified start-up and shutdown
What you need to get started
For the hosted route, the essentials are input images or video in a listed format, access to Replicate, and a decision about which outputs you want to inspect. A dedicated camera or GPU purchase is not required to try the hosted endpoint. If you choose local inference instead, plan for a developer setup and compute resources; the cited documentation does not provide a minimum hardware specification.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
