1. [Home](https://www.g2.com/) 2. ... 3. [Deep Learning Software](https://www.g2.com/categories/deep-learning) 4. [Image Recognition Software](https://www.g2.com/categories/image-recognition) 5. [Roboflow](https://www.g2.com/products/roboflow/reviews) 6. [Roboflow Claims vs Evidence](https://www.g2.com/products/roboflow/claims-vs-evidence) # Roboflow Claims vs Evidence ## Claim: “Run inference on-device without the headache of environment management, dependencies, managing CUDA versions, and more.” ##### Supported by reviews. Relevant reviewers report quick implementation on NVIDIA Jetson devices, clean APIs for edge deployment, and the ability to run locally or in the cloud. These reviews support on-device deployment with reduced setup friction, though they do not specifically mention CUDA versions or dependency management. - “Quick to implement on NVIDIA Jetson device, easy to train models” [Read Full Review](https://www.g2.com/survey_responses/12202942) - “Clean APIs for edge deployment and integration into more complex applications” [Read Full Review](https://www.g2.com/survey_responses/11496577) - “I appreciate the ability to run locally or on the cloud.” [Read Full Review](https://www.g2.com/survey_responses/11635986) Last updated Oct 5, 2026 Source: [https://roboflow.com/deploy](https://roboflow.com/deploy) ## Claim: “Run models directly on Roboflow’s infrastructure through an infinitely-scalable API.” ##### Mixed support from reviews. Reviewers provide direct support for running custom models through Roboflow’s hosted API and deploying them without managing infrastructure. However, other reviewers report low API rate limits and deployment costs that limit scalability, so the reviews support hosted API deployment but conflict with the claim of infinite scalability. - “I deploy my own custom models through Roboflow's serverless Hosted API, so I get a production endpoint without managing any infrastructure.” [Read Full Review](https://www.g2.com/survey_responses/13353598) - “The platform makes it fast and simple to annotate datasets, and training/deploying models is quick and straightforward. Overall, I’ve been very happy with the service and would recommend it to teams that want to go from data to production without a lot of friction.” [Read Full Review](https://www.g2.com/survey_responses/11631373) - “High cost of the subscribtion, API rate limit is low and not encouraging to be deployed in Enterprise/Commercial app.” [Read Full Review](https://www.g2.com/survey_responses/11654582) - “The single largest cost of being a roboflow user is model deployments, which has a significant impact on scalability leaving some solutions cost-prohibitive.” [Read Full Review](https://www.g2.com/survey_responses/11500328) Last updated Oct 5, 2026 Source: [https://roboflow.com/deploy](https://roboflow.com/deploy) ## Claim: “Switch to new model versions and architectures without adjusting your inference code.” ##### Supported by reviews. Relevant reviewers describe switching models as easy, supporting different model architectures across deployment environments, and benchmarking new models with a few clicks. These reviews support the claim’s practical direction, although they do not explicitly document whether inference code required no changes. - “It's easy to switch models” [Read Full Review](https://www.g2.com/survey_responses/11597256) - “Integration with different model architectures and the ability to deploy them in different environments (web, mobile, etc.) make it a complete and highly versatile solution.” [Read Full Review](https://www.g2.com/survey_responses/11631318) - “every new model such as YOLO or SAM appears on launch day so we can benchmark the upgrade on the same dataset with a few clicks.” [Read Full Review](https://www.g2.com/survey_responses/11496050) Last updated Oct 5, 2026 Source: [https://roboflow.com/deploy](https://roboflow.com/deploy) ## Claim: “Roboflow Inference: An open source and scalable way to run models on-device, with or without an internet connection.” ##### Mixed support from reviews. Reviewers report support for local, edge, and Jetson-based deployment, including use at scale, but another reviewer says the cost of self-hosted inference makes scaling deployments more expensive than anticipated. The reviews do not directly establish the open-source or offline aspects of the claim. - “I appreciate the ability to run locally or on the cloud.” [Read Full Review](https://www.g2.com/survey_responses/11635986) - “Clean APIs for edge deployment and integration into more complex applications” [Read Full Review](https://www.g2.com/survey_responses/11496577) - “Quick to implement on NVIDIA Jetson device, easy to train models, lot's of different training options, workflows with tons of tools to automate processes.” [Read Full Review](https://www.g2.com/survey_responses/12202942) - “The only drawback I’ve encountered is the cost associated with self-hosted inference, which makes scaling deployments more expensive than anticipated.” [Read Full Review](https://www.g2.com/survey_responses/11698300) Last updated Oct 5, 2026 Source: [https://roboflow.com/deploy](https://roboflow.com/deploy) ## Claim: “Deploy using a fully managed infrastructure with an API endpoint or on-device, internet connection optional” ##### Supported by reviews. Relevant reviewers report both a fully managed hosted API that provides a production endpoint without infrastructure management and deployment options on local or edge devices. These experiences support the claim's hosted and on-device deployment paths, though the reviews do not explicitly detail internet connectivity requirements. - “I deploy my own custom models through Roboflow's serverless Hosted API, so I get a production endpoint without managing any infrastructure.” [Read Full Review](https://www.g2.com/survey_responses/13353598) - “I appreciate the ability to run locally or on the cloud.” [Read Full Review](https://www.g2.com/survey_responses/11635986) - “Clean APIs for edge deployment and integration into more complex applications” [Read Full Review](https://www.g2.com/survey_responses/11496577) Last updated Oct 5, 2026 Source: [https://roboflow.com/deploy](https://roboflow.com/deploy) ## Claim: “Integrated workflow builder and deployment ensures what you configure runs in production.” ##### Supported by reviews. Relevant reviewers generally describe integrated workflows, reliable deployment, and moving models into production with little friction. One reviewer notes that downstream automation after human review still requires manual steps, which limits the claim’s universality. - “Roboflow has been a great fit for my production systems. I appreciate the wide choice of models and the reliability of their hosted inference—it just works.” [Read Full Review](https://www.g2.com/survey_responses/11631373) - “Roboflow makes it simple to label, train, and deploy vision models. The interface is clean and fast. You get prebuilt tools for dataset management, augmentations, and model training. The API lets you move models into production without custom pipelines.” [Read Full Review](https://www.g2.com/survey_responses/11653467) - “Roboflow has been central to our project building a roofing defect detection model. It makes dataset management and annotation efficient, and the workflows feature has allowed us to automate preprocessing, tagging, and uploading of images.” [Read Full Review](https://www.g2.com/survey_responses/11561989) - “The main challenge is automation after human review - exporting datasets and moving them downstream into our own applications still requires manual steps.” [Read Full Review](https://www.g2.com/survey_responses/11561989) Last updated Oct 5, 2026 Source: [https://roboflow.com/deploy](https://roboflow.com/deploy) ## Claim: “Monitor which devices are online/offline, inference volume by device, confidence metrics for each model, inference time, and see individual prediction results.” ##### Mixed support from reviews. Reviewers report robust monitoring and performance tracking, along with strong model metrics, supporting parts of the claim. However, another reviewer says that viewing and managing live-stream inference was difficult, indicating limitations in the monitoring experience. - “The model metrics are fantastic alongside the diagrams.” [Read Full Review](https://www.g2.com/survey_responses/12966972) - “Wish it was easier to view and manage live stream video inference with their Growth subscription.” [Read Full Review](https://www.g2.com/survey_responses/12202942) Last updated Oct 5, 2026 Source: [https://roboflow.com/deploy](https://roboflow.com/deploy) ## Claim: “Use roboflow.js to run models directly from the browser, allowing you to bring your model to the web.” ##### Supported by reviews. Relevant reviewers report deploying models to web environments and running inference from applications, which is consistent with bringing Roboflow models to the web. The reviews do not specifically mention roboflow.js, but they support the claimed web and app deployment capability. - “The speed with which I can annotate, train, and deploy models is incredible. Integration with different model architectures and the ability to deploy them in different environments (web, mobile, etc.) make it a complete and highly versatile solution.” [Read Full Review](https://www.g2.com/survey_responses/11631318) - “I like the pre-built Templates and pre-trained models on Roboflow Universe. Both let me immediately create a workflow and start running inference from my app.” [Read Full Review](https://www.g2.com/survey_responses/11675591) Last updated Oct 5, 2026 Source: [https://roboflow.com/deploy](https://roboflow.com/deploy)