OpenAI Releases Gpt-oss AI Model, Offers Bounty For Vulnerabilities

OpenAI have just released gpt-oss, an AI large language model (LLM) available for local download and offline use licensed under Apache 2.0, and optimized for efficiency on a variety of platforms without compromising performance. This is their first such “open” release, and it’s with a model whose features and capabilities compare favorably to some of their hosted services.

OpenAI have partnered with ollama for the launch which makes onboarding ridiculously easy. ollama is an open source, MIT-licensed project for installing and running local LLMs, but there’s no real tie-in to that platform. The models are available separately: gpt-oss-20b can run within 16 GB of memory, and the larger and more capable gpt-oss-120b requires 80 GB. OpenAI claims the smaller model is comparable to their own hosted o3-mini “reasoning” model, and the larger model outperforms it. Both support features like tool use (such as web browsing) and more.

LLMs that can be downloaded and used offline are nothing new, but a couple things make this model release a bit different from others. One is that while OpenAI have released open models such as Whisper (a highly capable speech-to-text model), this is actually the first LLM they have released in such a way.

The other notable thing is this release coincides with a bounty challenge for finding novel flaws and vulnerabilities in gpt-oss-20b. Does ruining such a model hold more appeal to you than running it? If so, good news because there’s a total of $500,000 to be disbursed. But there’s no time to waste; submissions need to be in by August 26th, 2025.

Brilliant Labs Has New Smart Glasses, With A New Display

Brilliant Labs have been making near-eye display platforms for some time now, and they are one of the few manufacturers making a point of focusing on an open and hacker-friendly approach to their devices. Halo is their newest smart glasses platform, currently in pre-order (299 USD) and boasting some nifty features, including a completely new approach to the display.

Development hardware for the Halo display. The actual production display is color, and integrated into the eyeglasses frame.

Halo is an evolution of the concept of a developer-friendly smart glasses platform intended to make experimentation (or modification) as accessible as possible. Compared to previous hardware, it has some additional sensors and an entirely new approach to the display element.

Whereas previous devices used a microdisplay and beam splitter embedded into a thick lens, Halo has a tiny display module that one looks up and into in the eyeglasses frame. The idea appears to be to provide the user with audio (bone-conduction speakers in the arms of the glasses) as well as a color “glanceable” display for visual data.

Some of you may remember Brilliant Labs’ Monocle, a transparent, self-contained, and wireless clip-on display designed with experimentation in mind. The next device was Frame, which put things into a “smart glasses” form factor, with added features and abilities.

Halo, being in pre-release, doesn’t have full SDK or hardware details shared yet. But given Brilliant Labs’ history of fantastic documentation for their hardware and software, we’re pretty confident Halo will get the same treatment. Want to know more but don’t wish to wait? Checking out the tutorials and documentation for the earlier devices should give you a pretty good idea of what to expect.

Numbers Station Simulator, Right In Your Browser

Do you find an odd comfort in the uncanny, regular intonations of a Numbers Station? Then check out [edent]’s numbers station project, which leverages the browser’s speech synthesis engine to deliver a ceaseless flow of (mostly) numbers, calmly-intoned in various languages.

The project is an entry for the annual JavaScript Golfing Competition, in which participants aim to create a cool program in 1024 bytes or less. It cleverly relies on the Web Speech API to deliver the speaking parts, which helps keep the code size tiny. The only thing it’s missing is an occasional shadow of static drifting across the audio.

If you’re new to numbers stations, our own [Al Williams] is here to tell you all about them. But there’s no need to tune into an actual mysterious radio signal just to experience weird numbers; just fire up [edent]’s project, put on some headphones, and relax if you can.

Engrave A Cylinder Without A Rotary Attachment? No Problem!

Laser-engraving a cylindrical object usually requires a rotary attachment, which is a motorized holder that rotates a cylindrical object in sync with the engraver. But [Samcraft] shows that engraving all around a mug can be done without a motorized rotary holder.

Separating a design into elements thin enough to engrave individually without losing focus is the key.

The basic idea is to split the design into a number of separate engraving jobs, each containing one element of the overall design, then setting the mug into a 3D printed jig and manually rotating it between jobs. To demonstrate, [Samcraft] selects a series of line-art flowers and plants which are ideal for this approach because there’s no need to minutely register the individual engravings with one another.

What about focus? [Samcraft] found that a design up to 45 mm wide could be engraved onto the curved surface of his mug before focus suffers too much. It’s true that this technique only works with certain types of designs — specifically those with individual elements that can be separated into tall and thin segments — but the results are pretty nice.

Laser engravers are a very serious potential eye hazard, and we are not delighted to see the way the shield around [Samcraft]’s engraver cannot close completely to accommodate the mug while the laser is active. But we’re going to assume [Samcraft] has appropriate precautions and eye protection in place off-camera, because laser radiation and eyeballs absolutely do not belong together, even indirectly.

Continue reading “Engrave A Cylinder Without A Rotary Attachment? No Problem!”

Zine Printing Tips From A Solopreneur

Zines (self-produced, small-circulation publications) are extremely DIY, and therefore punk- and hacker-adjacent by nature. While they can be made with nothing more than a home printer or photocopier, some might benefit from professional production while losing none of their core appeal. However, the professional print world has a few gotchas, and in true hacker spirit [Mabel Wynne] shares things she learned the hard way when printing her solo art zine.

As with assembling hardware kits, assembling a zine can take up a lot of physical table space.

[Mabel] says the most useful detail to nail down before even speaking to printers is the zine’s binding, because binding type can impact layout and design of an entire document. Her advice? Nail it down early, whether it’s a simple saddlestitch (staples through a v-shaped fold of sheets), spiral binding (which allows a document to lay flat), or something else.

Aside from paper and print method (which may be more or less important depending on the zine’s content) the other thing that’s important to consider is the finishing. Finishing consists of things like cutting, folding, and binding of the raw printed sheets. A printer will help arrange these, but it’s possible to do some or even all of these steps for oneself, which is not only more hands-on but reduces costs.

Do test runs, and prototype the end result in order to force unknown problems to the surface before they become design issues. Really, the fundamentals have a lot in common with designing and building kits or hardware. Check out [Mabel]’s article for the full details; she even talks a little about managing money and getting a zine onto shelves.

Zine making is the DIYer’s way to give ideas physical form and put them into peoples’ hands more or less directly, and there’s something wonderfully and inherently subversive about that concept. 2600 has its roots in print, but oddball disk magazines prove one doesn’t need paper to make a zine.

Reachy The Robot Gets A Mini (Kit) Version

Reachy Mini is a kit for a compact, open-source robot designed explicitly for AI experimentation and human interaction. The kit is available from Hugging Face, which is itself a repository and hosting service for machine learning models. Reachy seems to be one of their efforts at branching out from pure software.

Our guess is that some form of Stewart Platform handles the head movement.

Reachy Mini is intended as a development platform, allowing people to make and share models for different behaviors, hence the Hugging Face integration to make that easier. On the inside of the full version is a Raspberry Pi, and we suspect some form of Stewart Platform is responsible for the movement of the head. There’s also a cheaper (299 USD) “lite” version intended for tethered use, and a planned simulator to allow development and testing without access to a physical Reachy at all.

Reachy has a distinctive head and face, so if you’re thinking it looks familiar that’s probably because we first covered Reachy the humanoid robot as a project from Pollen Robotics (Hugging Face acquired Pollen Robotics in April 2025.)

The idea behind the smaller Reachy Mini seems to be to provide a platform to experiment with expressive human communication via cameras and audio, rather than to be the kind of robot that moves around and manipulates objects.

It’s still early in the project, so if you want to know more you can find a bit more information about Reachy Mini at Pollen’s site and you can see Reachy Mini move in a short video, embedded just below.

Continue reading “Reachy The Robot Gets A Mini (Kit) Version”

Mini Car Racing Game Really Shows Off Multicolor Printing

Quality 3D printing is a common hobbyist tool nowadays, and [wontonnn]’s mini arcade car racing game really shows off how 3D printing can bring parts from functional to fantastic. There are quite a few details we like in [wontonn]’s design, so let’s take a closer look.

The mini mechanical game is one of those treadmill-based car racing games in which the player navigates a little car between an onslaught of belt-borne obstacles. A little DC motor spins things up in a modular side assembly, and a hand-cranked option is available. The player’s car attaches via a magnet to a steering arm; if the player’s car gets knocked off the magnet, game over.

Treadmill belt segments print as large pre-assembled pieces, with ends that snap together without connectors. Belts like this are sometimes tricky, so this is worth keeping in mind should one ever need a similar part. Since there are no external fasteners or hardware to depend on, one could resize it easily to suit their own project purposes.

The finishing touches on the whole assembly look great. It used to be that the sort of colors and lettering seen here would come from a sticker or label, but [wontonn] gets clean lines and colors by raising (or sinking) different parts of the design. The checkerboard pattern, for example, has the light squares raised for printing in a different color.

Electromechanical arcade games have an appeal all their own, being a fusion of both mechanical and electric design that comes together in a special way. Want to make your own? Get inspired by the classic Lunar Lander reimagined, or check out this LEGO treadmill racer that takes an entirely different approach to the concept.

Continue reading “Mini Car Racing Game Really Shows Off Multicolor Printing”