Open Model Hosting & GPUs for Next.js: what 50 makers' products chose
Among 50 makers' products built with Next.js that we know the open model hosting & GPUs of, Modal leads: 19 of them (38%). Next come Hugging Face (15) and Baseten (8).
Per-token API or your own code on rented GPUs. Serverless model APIs host popular open models behind an OpenAI-compatible endpoint, so you pay per token and run nothing. Renting GPU compute lets you run any model or pipeline you like, but you package it, handle cold starts and pay for time, busy or idle.
What makers on Next.js chose
| Option | Next.js products | All makers |
|---|---|---|
| 19 (38%) | 32% | |
| 15 (30%) | 42% | |
| 8 (16%) | 12% | |
| 5 (10%) | 10% | |
| 3 (6%) | 6% | |
| 3 (6%) | 3% | |
| 2 (4%) | 3% | |
| 1 (2%) | 1% |
Live makers' products with a source for both: that they're built with Next.js, and what they use for open model hosting & GPUs. A product using two options counts in both; not market share. "All makers" is the option's share among every maker's product that made this decision.
Which to pick, by situation
- The open model you need is already served by a host and you want to pay per token
Together AI 3 here
Fireworks AI 2 here
Hugging Face 15 here - Your workload is your own Python code — a pipeline, batch job or custom model — that should scale to zero
Modal 19 here - You want a raw GPU machine for experiments, training or a long-running server
Runpod 5 here - One model is core to the product and needs steady latency under real traffic
Baseten 8 here
Hugging Face 15 here
Modal
Your code on GPUs19 Next.js products here (38%), against 32% of all makers' products.
Best forRunning your own Python code — inference, batch jobs, sandboxes — on GPUs defined in code, scaling to zero between calls.
Trade-offCold starts on large models, and your code is shaped around Modal's decorators.
“Serverless GPU infrastructure helps us scale to workloads of hundreds of H200s/B200s within seconds to maintain low latency for our retrieval models in a cost-effective way.”
“We use Modal as our sandbox provider for executing code within Lightfield chats.”
Open Model Hosting & GPUsMaker says so · source ↗“Solving inferencing for us at scale.”
Open Model Hosting & GPUsMaker says so · source ↗“Super easy to spin up and scale my API requests. Would highly recommend anyone using this.”
Open Model Hosting & GPUsMaker says so · source ↗“Modal helps us to deploy services effortlessly, and host backend infrastructure without any overheads. A beautiful DX.”
Open Model Hosting & GPUsMaker says so · source ↗“Modal gives us fast spin up times, simple serverless workflows, and a clean developer experience. It let us prototype Kodezi pipelines without managing GPUs or infra overhead. The combination of speed, scale, and low friction made it easier to iterate than the alternatives.”
Open Model Hosting & GPUsMaker says so · source ↗“Helps us run intensive AI models without worrying about cold starts or scaling.”
Open Model Hosting & GPUsMaker says so · source ↗“runs our GPU compute. fast cold starts and clean serverless primitives made the deploy layer possible”
Open Model Hosting & GPUsMaker says so · source ↗“Serverless GPU infrastructure helps us scale to workloads of hundreds of H200s/B200s within seconds to maintain low latency for our retrieval models in a cost-effective way.”
Open Model Hosting & GPUsMaker says so · source ↗“We run large training and inference jobs on Modal!”
Open Model Hosting & GPUsMaker says so · source ↗“Modal lets us run isolated, on-demand ingestion workers so large source imports stay fast and dependable.”
Open Model Hosting & GPUsMaker says so · source ↗
Hugging Face
Dedicated model deployments15 Next.js products here (30%), against 42% of all makers' products.
Best forDeploying almost any model from the Hugging Face Hub to a dedicated endpoint, or calling popular models through Inference Providers, which routes to partner hosts under one API.
Trade-offDedicated endpoints bill per hour while running; Inference Providers only covers models the partner hosts serve, at their prices.
“Our in-browser AI features use Hugging Face Transformers.js. Running ML models directly in the browser without a backend call that's the magic Hugging Face enables for local-first AI apps.”
“Hugging Face gave us access to specialized open models that complement proprietary LLMs for specific analysis tasks.”
Open Model Hosting & GPUsMaker says so · source ↗“Quanty utilizes Hugging Face's LLM to power its Knowledge Graph, ensuring precise and insightful market analysis.”
Open Model Hosting & GPUsMaker says so · source ↗“the model catalog behind RunInfra. every model users deploy comes from HF. nothing else comes close for open-source AI.”
Open Model Hosting & GPUsMaker says so · source ↗“Hugging Face is unmatched in fostering an open-source AI ecosystem. Its model hub provides access to thousands of pre-trained models, democratizing AI development. The platform’s ease of use, strong community, and robust API support make it a go-to for NLP, vision, and generative AI tasks.”
Open Model Hosting & GPUsMaker says so · source ↗“Great for sharing datasets!”
Open Model Hosting & GPUsMaker says so · source ↗“Our in-browser AI features use Hugging Face Transformers.js. Running ML models directly in the browser without a backend call that's the magic Hugging Face enables for local-first AI apps.”
Open Model Hosting & GPUsMaker says so · source ↗“Hugging Face provides an excellent collection of models that are compatible with Typeahead.”
Open Model Hosting & GPUsMaker says so · source ↗“Dataset hosting with Croissant metadata support”
Open Model Hosting & GPUsMaker says so · source ↗“Hugging Face made local embeddings practical for PMB. We wanted semantic and multilingual recall without requiring users to send their project memory to a hosted embedding API. The open model ecosystem lets PMB stay offline-first while still giving agents useful memory retrieval.”
Open Model Hosting & GPUsMaker says so · source ↗“The easiest way to play with open LLMs and embeddings.”
Open Model Hosting & GPUsMaker says so · source ↗
Baseten
Dedicated model deployments8 Next.js products here (16%), against 12% of all makers' products.
Best forDeploying a specific model as a production endpoint with autoscaling and monitoring.
Trade-offBuilt for production workloads; more setup than calling a shared serverless API.
“Incredible team that enables us to do truly amazing things everyday!”
“Incredible team that enables us to do truly amazing things everyday!”
Open Model Hosting & GPUsMaker says so · source ↗
Runpod
Your code on GPUs5 Next.js products here (10%), against 10% of all makers' products.
Best forRenting GPU machines by the hour or running serverless GPU workers, with control over the hardware.
Trade-offMore hands-on setup and operations than a managed platform.
“Runpod powers our AI video generation behind the scenes. Affordable GPU compute with great reliability and speed.”
“Runpod powers our AI video generation behind the scenes. Affordable GPU compute with great reliability and speed.”
Open Model Hosting & GPUsMaker says so · source ↗“Backend provider for GPU infra”
Open Model Hosting & GPUsMaker says so · source ↗“Super easy to use GPU infra that we use for inference and even training!”
Open Model Hosting & GPUsMaker says so · source ↗“RunPod has become indispensable for us developing AI applications with its easy to use serverless GPU workers. We don’t need to spend a lot of money renting dedicated GPUs in the development phase and testing out our apps with high volume traffic and reshaping our AI models into their perfect shape and Tada! 🎉 Our AI model is getting ready in days; not months!”
Open Model Hosting & GPUsMaker says so · source ↗
Together AI
Serverless model API3 Next.js products here (6%), against 6% of all makers' products.
Best forCalling many popular open models through one OpenAI-compatible API, with fine-tuning on the same platform.
Trade-offYou're limited to the models they host unless you pay for dedicated capacity.
“Together AI makes a wide variety of open source models accessible, at an industry leading cost. It lets us provide access to modules like Llama and a privacy-friendly versions of DeepSeek.”
“Together AI makes a wide variety of open source models accessible, at an industry leading cost. It lets us provide access to modules like Llama and a privacy-friendly versions of DeepSeek.”
Open Model Hosting & GPUsMaker says so · source ↗
Lambda
Your code on GPUs3 Next.js products here (6%), against 3% of all makers' products.
Best forHourly NVIDIA GPU instances and multi-GPU clusters with no egress fees, for training or self-hosted serving.
Trade-offMachines only — its serverless Inference API has been wound down, so packaging, serving and scaling are yours.
Also chosen, and not seen yet
Fewer Next.js products here: Fireworks AI (2), Morph (1).
Real options for open model hosting & GPUs not seen among these Next.js products yet: vLLM.
Getting open model hosting & GPUs right
If the model you need is already served as an API, use that — it's the least work and scales to zero. Rent GPUs only for custom models, fine-tunes a host won't serve, or non-LLM pipelines. Check cold-start times with real traffic before promising users response times.
- Leaving a dedicated endpoint or rented GPU running over a weekend of no traffic; hourly billing doesn't care whether anyone called it.
- Promising users sub-second responses before measuring cold starts on a large model under real traffic.
- Renting GPUs to serve a popular open model that a serverless API already hosts for less than the idle time would cost.
The rest of the Next.js stack
What makers' Next.js products chose for their other decisions, the most common pick for each:
- Hosting: Vercel 1,078 of 1,312
- Transactional Email: Resend 493 of 830
- Database: Supabase 171 of 382
- Payments: Stripe 288 of 359
- LLM API: OpenAI API 222 of 301
- Authentication: Clerk 176 of 211
- Web Analytics: Google Analytics 105 of 197
- Product Analytics: PostHog 123 of 168
Questions
What do most Next.js apps use for open model hosting & GPUs?
Among 50 makers' products built with Next.js that we know the open model hosting & GPUs of, Modal leads: 19 of them (38%). Next come Hugging Face (15) and Baseten (8).
How this is counted: a product is on Next.js when a live-product source says so (the maker, its website, a customer story or a subprocessor list), and the same for its open model hosting & GPUs. How evidence is collected.