Run Models Locally for Next.js: what 15 makers' products chose
Among 15 makers' products built with Next.js that we know the run models locally of, Ollama leads: 9 of them (60%). Next come Llama (5) and llama.cpp (1).
App or engine. Desktop apps and CLIs download models for you and expose a local API in minutes. The underlying engine gives you full control over builds, quantization and flags, and you handle the details yourself.
What makers on Next.js chose
| Option | Next.js products | All makers |
|---|---|---|
| 9 (60%) | 64% | |
| 5 (33%) | 34% | |
| 1 (7%) | 2% |
Live makers' products with a source for both: that they're built with Next.js, and what they use for run models locally. 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
- You want a local model behind an API for your code and dev tools in minutes
Ollama 9 here - You ship local inference inside your own app or need settings the apps don't expose
llama.cpp 1 here
Ollama
Local runtime (CLI)9 Next.js products here (60%).
Best forPulling and running models with one command, with a local API that other tools already support.
Trade-offFewer tuning knobs than running the engine directly.
“Ollama helped us run local models, evals, all without thinking of overhead”
“Universal connector for all the local models hooking into our SDK.”
Run Models LocallyMaker says so · source ↗“Ollama helped us run local models, evals, all without thinking of overhead”
Run Models LocallyMaker says so · source ↗“Allowing us to run on device models for the Superapp”
Run Models LocallyMaker says so · source ↗“Ollama powers our local LLM commands users can control the video editor in plain English like "remove the intro" or "speed up the middle." It runs entirely on the user's machine with no API keys needed. Ollama made local AI feel as easy as running a Docker container.”
Run Models LocallyMaker says so · source ↗“By embedding Ollama directly in the app my users can communicate with local AI models with ease without having to rely on a 3rd party API”
Run Models LocallyMaker says so · source ↗“Powers the local AI assistant (Nibby) without sending family data to third-party APIs. Privacy-first approach — everything runs on your own machine.”
Run Models LocallyMaker says so · source ↗“Ollama is the choice for our users wanting local graphs, works best with 32b param models.”
Run Models LocallyMaker says so · source ↗“Ollama's the recommended pick because it's local, free, no keys; fits the fully-local thesis. Z.ai isn't one of the 10 wired-in providers, though it might slot in via the openai-compatible option, untested.”
Run Models LocallyMaker says so · source ↗“It's one of the best apps to run models locally”
Run Models LocallyMaker says so · source ↗
Llama
Open-weight model5 Next.js products here (33%).
Best forMeta's widely supported open-weight models, available in every local runtime and in many sizes.
Trade-offCustom license with use restrictions rather than an open-source license; no new Llama release since Llama 4 in April 2025, as Meta's developer platform now centres on its hosted Muse models.
“Llama 3: Llama 3 has empowered our project with its powerful, open-source capabilities, offering flexibility and innovation in agent development.”
“Llama 3: Llama 3 has empowered our project with its powerful, open-source capabilities, offering flexibility and innovation in agent development.”
Run Models LocallyMaker says so · source ↗“Llama's open source models allow us to run LLM's on an otherwise infeasible scale of data”
Run Models LocallyMaker says so · source ↗“LLama 4 Scout, a great model for simple tasks!”
Run Models LocallyMaker says so · source ↗“Big thanks to Meta and the Llama 3 team for their contributions to the open-source AI landscape. The availability of powerful, openly accessible models like Llama 3 is driving incredible innovation across the industry. By integrating open-source models into our platform, we can offer unparalleled versatility and empower our users to leverage the best tools available. Together, we're advancing AI in ways that prioritize openness, accessibility, and powerful functionality.”
Run Models LocallyMaker says so · source ↗“We use Llama as part of Userology’s research engine for fast, structured UX work like tagging, clustering, and pattern-finding in sessions. Because it’s open-weight, we get tighter control over performance, routing, and data privacy than most closed models, which makes our multi-model stack more flexible for different customer needs.”
Run Models LocallyMaker says so · source ↗Also chosen, and not seen yet
Fewer Next.js products here: llama.cpp (1).
Real options for run models locally not seen among these Next.js products yet: LM Studio, Jan, MLX LM, Qwen, gpt-oss, Gemma.
Getting run models locally right
Start with a small quantized model (a few billion parameters) and check speed on your actual machine before planning a feature around it. Point your code at the local OpenAI-compatible endpoint so switching to a hosted model later is a config change. Expect a big gap in quality and speed between a laptop and hosted frontier models; test on your real task.
- Choosing a model by its benchmark scores and only then finding it runs at two tokens a second on your machine.
- Assuming "open weights" means open source; some model licenses restrict commercial use, user counts or what you can build.
- Planning a local-first feature around your own high-end laptop when many users have far less memory.
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 run models locally?
Among 15 makers' products built with Next.js that we know the run models locally of, Ollama leads: 9 of them (60%). Next come Llama (5) and llama.cpp (1).
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 run models locally. How evidence is collected.