{"id":594901,"date":"2026-08-24T22:58:21","date_gmt":"2026-08-24T22:58:21","guid":{"rendered":"https:\/\/www.newsbeep.com\/il\/594901\/"},"modified":"2026-08-24T22:58:21","modified_gmt":"2026-08-24T22:58:21","slug":"jetbrains-junie-now-runs-entirely-offline-can-you-spare-a-64-gb-m5-mac","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/il\/594901\/","title":{"rendered":"JetBrains Junie now runs entirely offline. Can you spare a 64 GB M5 Mac?"},"content":{"rendered":"

While most AI coding tools default to cloud-hosted models, local model runtimes have become a viable alternative<\/a> for developers who want to keep code on their own machines, avoid per-request API costs, or have to work without an internet connection. <\/p>\n

Tools such as Cline<\/a>, Continue<\/a>, and Aider<\/a> can already be pointed at runtimes like Ollama<\/a> or LM Studio<\/a>. At the same time, GitHub added local-model support to Copilot CLI in April<\/a>, including an offline mode for fully air-gapped setups.<\/p>\n

The catch is that \u201clocal\u201d generally still leaves much of the assembly to the developer. You have to choose a model and a suitable quantization for your hardware, configure the runtime and context settings, and work out which combination performs well with the agent. <\/p>\n

And that last part really does matter: models small enough to run comfortably on a laptop can still struggle with the tool use, reasoning, and longer-running tasks that coding agents demand.<\/p>\n

This is why JetBrains has now built Junie Local<\/a>, a free version of its coding agent designed to run entirely on the developer\u2019s own machine.<\/p>\n

Local market<\/p>\n

By way of a brief recap, JetBrains<\/a> \u2014 the developer tools company behind IntelliJ IDEA<\/a>, PyCharm<\/a> and WebStorm<\/a> \u2014 launched Junie in January 2025<\/a> as an AI coding agent embedded in its IDEs, capable of planning tasks, modifying code, running tests and inspections, and working with the context of a developer\u2019s project. It has since expanded into a standalone CLI<\/a>.<\/p>\n

Junie itself isn\u2019t exactly new to local models. In a blog post<\/a> published on Monday, JetBrains\u2019 head of marketing Dmitry Savelev<\/a> notes that developers have been able to connect the agent to runtimes such as Ollama and LM Studio for some time, load whichever model they want, and have Junie run against it locally.<\/p>\n

However, with Junie Local, JetBrains has picked the model, quantized it, and tuned its inference engine and agent harness around that specific combination. Setup is handled from inside Junie itself: running \/local downloads the model and inference engine, starts a local server, and switches the agent over automatically. There is no separate Ollama or LM Studio installation, endpoint to configure, or model profile to write.<\/p>\n

The first step is simply choosing Junie Local from the model selector, where it appears alongside the usual array of cloud-hosted models.<\/p>\n

\"Junie\u2019sJunie\u2019s model selector offers Junie Local alongside its cloud-hosted models<\/p>\n

Once the download and setup are complete, Junie switches to the local Qwen model, which then appears in the CLI like any other model option.<\/p>\n

\"JunieJunie running with Qwen3.6 locally<\/p>\n

From that point on, inference happens entirely on the developer\u2019s machine.<\/p>\n

Under the hood: Why Qwen3.6 \u2014 and why an M5 Mac<\/p>\n

It\u2019s worth noting that JetBrains has been very specific about its model choice and hasn\u2019t opted for the latest, shiniest open-weight version. Junie Local uses Qwen3.6-27B, a 27-billion-parameter open-weight model released in April<\/a>, even though the newer Qwen3.8-<\/a>2<\/a>7B<\/a> arrived earlier in August with improvements.<\/p>\n

\u201cOn today\u2019s Macs, [Qwen] 3.6 wins.\u201d<\/p>\n

Savelev notes that the choice came down to how the two models behaved inside Junie on current Macs, with Qwen3.8 requiring its reasoning mode to be enabled to work reliably with the agent; with reasoning switched on, tasks took roughly four times longer. For Junie Local right now, Qwen 3.6 offers the better balance of reliability and speed.<\/p>\n

\u201cOn today\u2019s Macs, 3.6 wins,\u201d Savelev writes. <\/p>\n

JetBrains runs Qwen3.6-27B at 4-bit using an inference engine based on mlx-vlm, which in turn uses MLX<\/a>, Apple\u2019s machine-learning framework for Apple Silicon. It\u2019s a similar underlying approach to the one Ollama adopted in March<\/a>, when it moved its Apple Silicon engine onto MLX to take advantage of the chips\u2019 unified-memory architecture.<\/p>\n

There is a fairly substantial hardware floor, though: JetBrains confirms that Junie Local involves about 20 GB of downloads, and requires macOS 26, at least 64 GB of unified memory, and an Apple M5 chip or newer. In real terms, that 64 GB requirement puts MacBook Pro users into M5 Pro or M5 Max territory \u2014 in other words, this is firmly a high-end Mac proposition.<\/p>\n

JetBrains acknowledges that those requirements will put Junie Local beyond the reach of plenty of developers who might otherwise be interested in running it.<\/p>\n

\u201cWe know that an M5 Mac with 64 GB of RAM is a big ask,\u201d Savelev writes. \u201cWe are not going to pretend otherwise. That is simply what it costs to run a 27B model well today, and it is the number we are working hardest to bring down.\u201d<\/p>\n

\u201cWe know that an M5 Mac with 64 GB of RAM is a big ask. We are not going to pretend otherwise.\u201d<\/p>\n

The intention is to reduce memory requirements, support a wider range of hardware, and continue optimizing the underlying stack.<\/p>\n

\u201cIf the lofty requirements are the reason you cannot try Junie Local, rest assured that we are working to bring them down,\u201d Savelev adds.<\/p>\n

Where local pays off<\/p>\n

Ultimately, the hardware requirement is closely tied to where JetBrains identifies the real performance bottleneck for a local coding agent. The tokens-per-second metric measures how quickly a model generates output, but an agent can spend much of its time first ingesting source files, prompts and other context \u2014 the prefill stage \u2014 before it starts producing an answer.<\/p>\n

\u201cEveryone benchmarks generation speed,\u201d Savelev writes. \u201cFor a coding agent, that turns out to be the wrong number to chase because most of the time is spent on prefill, while the model reads files to work out what is going on. Optimizing for prefill is where the real gains were.\u201d<\/p>\n

Being free and unmetered also changes the kinds of jobs developers might be willing to hand over. JetBrains positions Junie Local as particularly well-suited to long, repetitive, and mechanical work \u2014 multi-file refactors and renames, filling test-coverage gaps, dependency upgrades, and framework migrations \u2014 where the agent can keep working and iterating without the developer having to think about how many tokens it\u2019s burning through.<\/p>\n

\u201cLong, repetitive, mechanical work is exactly what an agent is for, and exactly what you stop asking for when you are keeping an eye on your balance,\u201d Savelev writes.<\/p>\n

\u201cLong, repetitive, mechanical work is exactly what an agent is for, and exactly what you stop asking for when you are keeping an eye on your balance.\u201d<\/p>\n

For everyday development work, Savelev reckons users are unlikely to notice much of a gap compared with stronger cloud models. However, he does concede that more complex architectural reasoning remains better suited to those models.<\/p>\n

And then, of course, there is arguably the biggest reason developers have been interested in local models in the first place: privacy. Running the entire agent locally means that no external model provider sits between the developer and their code, and that no source, prompts, or generated changes need to leave the machine. For developers working on proprietary code, under client NDAs, or in environments where sending source to a third party is simply off the table, that is a substantial part of the appeal.<\/p>\n

\u201cEverything after the download happens on your hardware, so your prompts, source, and diffs stay put,\u201d Savelev writes.<\/p>\n

\t<\/p>\n

\n\t\t\t\tYOUTUBE.COM\/THENEWSTACK\n\t\t\t<\/p>\n

\n\t\t\t\tTech moves fast, don’t miss an episode. Subscribe to our YouTube
\n\t\t\t\tchannel to stream all our podcasts, interviews, demos, and more.\n\t\t\t<\/p>\n

\t\t\t\tSUBSCRIBE<\/p>\n

\t<\/a><\/p>\n

Group
\n Created with Sketch.<\/p>\n

\t\t<\/p>\n

\t\t\t\t\t<\/p>\n

\n\t\t\t\t\t\t\tPaul is an experienced technology journalist covering some of the biggest stories from Europe and beyond, most recently at TechCrunch where he covered startups, enterprise, Big Tech, infrastructure, open source, AI, regulation, and more. Based in London, these days Paul…\t\t\t\t\t\t<\/p>\n

\t\t\t\t\t\tRead more from Paul Sawers\t\t\t\t\t\t<\/p>\n

\t\t<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"While most AI coding tools default to cloud-hosted models, local model runtimes have become a viable alternative for…\n","protected":false},"author":2,"featured_media":594902,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[6],"tags":[85,46,125],"class_list":["post-594901","post","type-post","status-publish","format-standard","has-post-thumbnail","category-technology","tag-il","tag-israel","tag-technology"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts\/594901","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/comments?post=594901"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts\/594901\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/media\/594902"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/media?parent=594901"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/categories?post=594901"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/tags?post=594901"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}