# What local AI coding model can a contractor run on a Mac Studio or DGX Spark?

[Recursive Flint](/tools/recursive-flint), a new open-weights coding model released Oct 21, 2026, is built to run on your own hardware. Recursive says it is a mixture-of-experts model with 40B total and about 2.5B active parameters. It's free under the Apache 2.0 license and published on Hugging Face in two local builds: MLX INT4 weights for Apple Silicon and an NVFP4 GGUF for NVIDIA.

Mac Studio: the INT4 Mac build is about 22.6 GB, so it fits a Mac Studio with 48 GB or more of memory. WIRED's hands-on was on a 48 GB Mac. DGX Spark: the GGUF model card recommends 48 GB of memory, and the [DGX Spark](/tools/nvidia-dgx-spark-hardware) we list has 128 GB of unified memory, shared with DGX OS. Recursive says Flint runs at 125+ tokens per second on an M4 Pro MacBook Pro and 150+ on a 64 GB RTX Spark at short contexts. Those are Recursive's own numbers, and so are all of its benchmark scores. We haven't seen independent tests yet, or speeds at long contexts.

What it's for: Flint is a coding model. Use it to build small tools, scripts and spreadsheet automations for your shop, not to write bids or proposals. Recursive points Mac users to its own Rec launcher and [llama.cpp](/tools/llama-cpp). [oMLX](/tools/omlx), the MLX model server for Macs, merged native Flint support on Oct 23, 2026. It's in oMLX's 0.7.3 development builds (v0.7.3.dev1, Oct 24, and later), but not yet in a stable release.

DIY-first approach: download the build for your machine from Recursive's Hugging Face page, try it on one real task like a script that renames job photos, and check the output before you rely on it. For setup help see our [Mac Studio guide](/guides/local-ai-mac-studio-mlx), [DGX Spark setup guide](/guides/dgx-spark-setup) and [model and hardware fit guide](/guides/local-ai-models-hardware-fit), plus the [Mac Studio](/tools/mac-studio-hardware) listing and our [local AI category](/category/local-ai).

> **What we run.** For day-to-day local work in our contracting shop, we run **DeepSeek V4 Flash** on **oMLX** on a **Mac Studio with an M3 Ultra chip and 512 GB of unified memory**. oMLX is a free, open-source MLX server for Apple Silicon. Recursive Flint is a different kind of tool: an open-weights model built specifically for coding, aimed at building small scripts and shop tools on your own machine. Recursive's own Rec engine lists M4 Pro and M5 Pro/Max chips, not the M3 Ultra, so Recursive's speed figures may not match a machine like ours. oMLX added native Flint support in its 0.7.3 development builds, not yet a stable release, so check that your installed version includes it. **We haven't timed Flint ourselves; the speeds and scores on this page are Recursive's own.** (DeepSeek note: federal civilian agencies are barred from DeepSeek products; [see our FAQ](/questions/can-contractors-still-run-deepseek-locally).)

*AI-assisted, reviewed by our team. Last verified 2026-10-26.*

## Sources

- [Recursive: Introducing Recursive Flint](https://recursive.com/updates/introducing-recursive-flint)
- [Hugging Face: RecursiveSI/flint](https://huggingface.co/RecursiveSI/flint)
- [Hugging Face: RecursiveSI/flint-int4 (MLX INT4, Apple Silicon)](https://huggingface.co/RecursiveSI/flint-int4)
- [Hugging Face: RecursiveSI/flint-nvfp4-gguf (NVIDIA)](https://huggingface.co/RecursiveSI/flint-nvfp4-gguf)
- [WIRED: Flint hands-on (Oct 21, 2026)](https://www.wired.com/story/recursive-self-improvement-flint-latest-model-scary-good/)
- [TechCrunch: Recursive launches Flint (Oct 21, 2026)](https://techcrunch.com/2026/10/21/recursive-launches-flint-a-laptop-coding-model-it-says-was-largely-built-by-ai/)
- [GitHub: oMLX pull request #4763, Add native Flint serving (merged Oct 23, 2026)](https://github.com/jundot/omlx/pull/4763)
- [oMLX v0.7.3.dev1 release notes (GitHub)](https://github.com/jundot/omlx/releases/tag/v0.7.3.dev1)

## Related

- [Recursive Flint](https://contractoraishop.com/tools/recursive-flint)
- [oMLX](https://contractoraishop.com/tools/omlx)
- [Mac Studio (hardware)](https://contractoraishop.com/tools/mac-studio-hardware)
- [NVIDIA DGX Spark (hardware)](https://contractoraishop.com/tools/nvidia-dgx-spark-hardware)
- [llama.cpp](https://contractoraishop.com/tools/llama-cpp)
- [local-ai-mac-studio-mlx](https://contractoraishop.com/guides/local-ai-mac-studio-mlx)
- [dgx-spark-setup](https://contractoraishop.com/guides/dgx-spark-setup)
- [local-ai-models-hardware-fit](https://contractoraishop.com/guides/local-ai-models-hardware-fit)

HTML: https://contractoraishop.com/questions/local-ai-coding-model-mac-studio-dgx-spark