DeepSeek V4 Flash 0731

DeepSeek

Free

Get

MLX Score3.7

PriceFreeno paid plan

RunsOn your Mac

DeveloperDeepSeek

LicenceOpenMIT

CategoryModels

Mac14.0+macOS or later

DeepSeek V4 Flash 0731 is the official release of DeepSeek-V4-Flash, with stronger agentic abilities than the preview. It's a mixture-of-experts model with 284 billion parameters, 13 billion active per token, and a context of 1 million tokens.

A 4-bit MLX build is about 151 GB; on mlx.app's estimate it needs a Mac with 256 GB of memory. Stock mlx-lm can't load it yet: support for its architecture is in an unmerged pull request (ml-explore/mlx-lm#1192), which oMLX ships as a patch.

Open weights under MIT. Run locally, nothing leaves your Mac.

How to Get It

Run it on your MacSee how

WebsiteVisit

Official downloads only. MLX never re-hosts an app, and never runs a command for you.

MLX Score

3.7out of 5

Hugging Face downloads and likesahead of 67%

Hugging Face: 3,029,525 downloads (30 days), 4,094 likes · read 2026-11-02

Ranked against the open models listings, from real public data only. How the score works.

Information

Developer: DeepSeek
Category: Open models
Price: Free (Open weights, free to download.)
Your data: Runs on your Mac
Licence: MIT
Mac: macOS 14.0 or later
iPhone and iPad: —
Needs: 256 GB memory
Links: Website · Hugging Face
Facts checked: 2026-11-01

Sources

huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731 · MIT, official release of V4-Flash, agentic capabilities · checked 2026-11-01huggingface.co/mlx-community/DeepSeek-V4-Flash-4bit · 4-bit MLX build (about 151 GB, converted from the April preview) · checked 2026-11-01ml-explore.github.io/mlx/build/html/install.html · macOS 14.0 minimum (MLX needs Apple silicon and macOS 14.0+) · checked 2026-11-01mlx.app/run/deepseek-v4-flash-0731 · memory: smallest Mac (256 GB) where the 4-bit MLX build fits with room for context, from lib/compare (75% of unified memory usable) · checked 2026-11-01huggingface.co/deepseek-ai/DeepSeek-V4-Flash · V4-Flash: 284B total / 13B activated, 1M context (0731 has the same structure per its card) · checked 2026-11-01github.com/ml-explore/mlx-lm/pull/1192 · deepseek_v4 architecture not yet in mlx-lm (unmerged PR), per lib/compare/models.source.json · checked 2026-11-01Icon from Hugging Face

See All