Strata review

Q: How do I install Strata?

Run curl -fsSL https://github.com/Niko1221/Strata/releases/latest/download/install.sh -o install.sh && sh install.sh. The install section above has the other options from the project's README.

Strata is a local inference engine and app that runs large open models like Qwen3.8-Flash-Next on consumer hardware with OpenAI/Anthropic APIs.

Updated Nov 1, 2026. Review written Nov 1, 2026 from the project's own sources; usage data refreshed every Monday.

TL;DR

Strata runs models such as Qwen3.8-Flash-Next, Qwen3.6-35B-A3B and DeepSeekV4.1-Flash on your own computer. Strata is for developers who want local chat and coding without an account. It offers a desktop app, a terminal app, and OpenAI- and Anthropic-compatible APIs on localhost for tools like Claude Code and Codex CLI.

37.1kGitHub stars
+5.1knew stars, week of Oct 25
#3in Inference and training this week

Strata at a glance

Category: Inference and training
Maintained by: Niko Veit (individual, @Niko1221 on GitHub)
License: MIT
Language: C++, Python, Cuda
Repository: github.com/Niko1221/Strata
Latest release: v0.2.9 on GitHub, Oct 31, 2026
First published: Sep 24, 2026

What Strata does

Strata runs Qwen3.8-Flash-Next, Qwen3.6-35B-A3B and DeepSeekV4.1-Flash locally, with a choice of quantized model packs from several providers.
Strata serves OpenAI- and Anthropic-compatible APIs on localhost, with optional image input.
Strata connects to coding tools such as Claude Code, Codex CLI, OpenCode and Hermes Agent through its compatible APIs.
Strata's desktop app combines chat, files and a live preview in one window.
Strata reads DeepSeek model weights that do not fit in memory from your SSD.
Strata's terminal app lets you watch your model and hardware while it runs.

How it works

Strata runs as a desktop app or a terminal app. On first start the desktop app downloads the engine for your graphics card. In the terminal, strata launch prompts you to download and start a model. The engine exposes OpenAI- and Anthropic-compatible APIs on localhost, and strata connect shows setup instructions for coding tools. Docker and building from source are also documented.

How to install and start using Strata

Linux / Mac

curl -fsSL https://github.com/Niko1221/Strata/releases/latest/download/install.sh -o install.sh && sh install.sh

Windows PowerShell

irm https://github.com/Niko1221/Strata/releases/latest/download/install.ps1 | iex

Requirements

A supported GPU or system: NVIDIA RTX 20–50 series, supported AMD Radeon cards, Intel Arc on Linux, Apple Silicon M1 or newer, Strix Halo, or DGX Spark.
DeepSeekV4.1-Flash needs about 245 to 280 GB of free disk space, depending on the set.
DeepSeekV4.1-Flash on Apple Silicon needs a Mac with 64 GB or more and is text only.
Strata needs no account.

First steps

Install the terminal app. On Windows, open a new terminal afterwards.
Run strata launch and follow the prompts to download and start a model.
Run strata connect to see setup instructions for your coding tool.

Who Strata is for

Developers who want to run large open models on their own consumer GPU.
Users of Claude Code, Codex CLI or OpenCode who want a local model backend.
Developers who want a local, OpenAI- or Anthropic-compatible API endpoint.

Limitations to know

Strata's AMD Radeon RX 6800 / 6900 support is experimental, and Intel Arc and DGX Spark support is Linux only.
Strata's DeepSeek sets are 239 GB or larger to download and rely on SSD reads when memory runs short.
Strata's Gyro model packs use less memory but run slower.

Is Strata growing? Usage data

In the week of Oct 25–31, 2026, Strata ranked #4 on the Devtoolwatch Board and #3 in Inference and training, with a momentum score of 98 out of 100.

New GitHub stars per week for the last 12 weeks, from GitHub's stargazer history (today's stargazers by the week they starred).

Show the weekly numbers

New GitHub stars per week for Strata
Week of	New stars
Aug 9, 2026	–
Aug 16, 2026	–
Aug 23, 2026	–
Aug 30, 2026	–
Sep 6, 2026	–
Sep 13, 2026	–
Sep 20, 2026	217
Sep 27, 2026	8.5k
Oct 4, 2026	12k
Oct 11, 2026	6.1k
Oct 18, 2026	5.2k
Oct 25, 2026	5.1k

On Hacker News

Run Qwen 3.8 Flash Next (125B) on consumer hardware (RTX 4090) at 100T/s, 940 points and 420 comments (Oct 4, 2026)
Strata: Qwen3.8-Flash-Next (125B Moe) on a 8GB+ Nvidia GPU, 5 points and 0 comments (Sep 28, 2026)

The Board ranks tools by four signals each week: new stars, new stars for the repo's size, added downloads and download growth beyond the market. How the ranking works.

Strata alternatives

Other LLM inference and training tools on the Board in the week of Oct 25–31, 2026, by momentum:

Inference and training ranked by momentum, the week of Oct 25–31, 2026
Rank	Tool	Stars	Downloads/wk
#1	dsh-our-free-modeldsh-our-free-model is a dsh plugin that provides DeepSeek V4.1 Flash, Kimi K3 and other models without login, sign-up or API key.	54.3k	–
#2	magpiemagpie is a local gateway and app that lets coding agents like Claude Code, Codex and Gemini CLI use any provider's model.	26.9k	–
#4	edge0edge0 is an open-source streaming inference framework that runs large sparse mixture-of-experts models on consumer hardware by streaming…	8.9k	–
#5	OmniRouteOmniRoute is an open-source AI gateway that routes requests across many model providers through one OpenAI-compatible endpoint, with…	83.2k	106k
#6	CLIProxyAPICLIProxyAPI is a Go proxy server that exposes CLI coding accounts like Codex, Claude Code and Gemini CLI through OpenAI, Gemini and…	57.1k	–
#7	codex-chatgpt-webcodex-chatgpt-web is a launcher that lets Codex use the ChatGPT Web models on your account, including Pro, through browser automation.	15.3k	–

All LLM inference and training tools, ranked

Recent Strata releases

v0.2.9, Oct 31, 2026
v0.2.8, Oct 29, 2026
v0.2.7, Oct 27, 2026
v0.2.6, Oct 25, 2026
v0.2.5, Oct 25, 2026

Strata FAQ

Is Strata free and open source?

Yes. Strata's source code is on GitHub (Niko1221/Strata) under the MIT license, so you can use it for free.

How do I install Strata?

Run curl -fsSL https://github.com/Niko1221/Strata/releases/latest/download/install.sh -o install.sh && sh install.sh. The install section above has the other options from the project's README.

Is Strata still maintained?

Yes. The Niko1221/Strata repository was last updated on Oct 31, 2026, and the latest release, v0.2.9 on GitHub, came out on Oct 31, 2026.

How popular is Strata?

Strata has 37.1k GitHub stars as of Nov 1, 2026. It gained 5.1k new stars in the week of Oct 25–31, 2026, against a 12-week average of 6.2k a week. That made it #3 among LLM inference and training tools on the Devtoolwatch Board.

What are the best alternatives to Strata?

Other LLM inference and training tools on the Devtoolwatch Board in the week of Oct 25–31, 2026, by momentum: dsh-our-free-model (#1), magpie (#2), edge0 (#4), OmniRoute (#5), CLIProxyAPI (#6). The alternatives table above has their numbers.

Which coding tools work with Strata?

Strata works with Claude Code, Codex CLI, OpenCode, oh-my-pi, pi.dev, Hermes Agent, OpenClaw, and other tools through its OpenAI- and Anthropic-compatible APIs. Run strata connect to see setup instructions for your tool.

Does Strata need an account or an internet API?

Strata needs no account. Models run on your own computer, and Strata serves its APIs on localhost.

Which model pack should I pick in Strata?

Strata recommends ISTA-DASLab packs first, then Unsloth and other providers where a compatible model fits. For coding, the README points to a Coder IQ1_M pack. A guide in docs/MODELS.md helps you choose by memory.

Can Strata run DeepSeekV4.1-Flash on a Mac?

Strata runs DeepSeekV4.1-Flash on Apple Silicon Macs with 64 GB or more, text only. Support was added in release 0.2.7.1.

Does Strata accept images?

Strata supports optional image input.

Sources

Strata README badge

Maintainers can show the tool's weekly Board rank in their README:

[![Devtoolwatch](https://index.devtoolwatch.com/badge/niko1221-strata.svg)](https://index.devtoolwatch.com/tools/niko1221-strata/)

How this page is made: Devtoolwatch writes each review from the project's README, releases, package metadata and website, with an AI model, then checks every claim against the source it cites and drops what the sources don't support. The numbers come from Devtoolwatch's daily GitHub snapshots, GitHub's stargazer history, and the npm and PyPI download APIs. Spotted an error? Email [email protected].