Open Models news

Runway opens its Praxis-1 robot model to everyone

The video AI company released the weights and code of its robot model under the Apache 2.0 licence. Runway says it beats FLUX 3 Action on NVIDIA’s RoboLab-120 test. Runway ran the tests itself, and the model ranks fifth on the test’s overall list.

Runway released Praxis-1 with open weights on November 11. The 27-billion-parameter model learns from video how objects move, then turns camera images and text instructions into robot movements. Runway first showed it in September, but only to selected partners.

Thinking Machines launches Inkling 2 Small, weights to come

The model scores 31 on Artificial Analysis’s index, level with Gemini 3.8 Flash (high). It would be the top open model from outside China, but its weights are not public yet. Thinking Machines plans them for next week, after safety tests.

Thinking Machines Lab, founded by former OpenAI technology chief Mira Murati, released Inkling 2 Small on November 5. For now it runs only on the company’s Tinker service. The company plans to publish the weights under an Apache 2.0 licence once outside safety testing ends.

Kimi K3.1 Pro’s weights are out, and Tencent launches Hy4

Moonshot’s 2.8-trillion-parameter model can now be downloaded, under a custom licence with limits for the biggest businesses. Tencent’s Hy4 adds vision and computer use, but its weights are only promised “in the coming days”.

Moonshot AI has published the full weights of Kimi K3.1 Pro on Hugging Face, keeping its promise to do so by November 4. Anyone can download them without asking. On November 4, Tencent launched Hy4 in its apps and through its API. Tencent says Hy4’s weights will follow soon, so for now only the August preview can be downloaded.

Mistral releases the weights of its Large 4 model

Anyone can now download the 1-trillion-parameter model, under Mistral’s own licence. Mistral says it ties China’s GLM-5.3 at coding, but Artificial Analysis scores it one point lower there and further behind overall.

French company Mistral AI published the weights of Mistral Large 4 on Hugging Face on October 29, three weeks after a public preview. The model, nicknamed “Le Chonk”, has about 1 trillion parameters, and about 50 billion of them work on each token. Its API costs $0.70 per million input tokens and $2.10 per million output tokens.

Nvidia’s open Nemotron model transcribes 40 languages

Nemotron 3.5 Transcribe turns recordings into text, translates and labels who is speaking. Anyone can download it for commercial use, and Nvidia also released labels for 463,062 hours of speech.

Nvidia released Nemotron 3.5 Transcribe on Hugging Face on October 28. It is an open-weight model that writes down speech in 40 languages, and Nvidia says it translates 31 of them to or from English. It can tell up to eight speakers apart. Nvidia also published Granary v2, a large set of labels for training speech models.

Opera lets websites use AI that runs on your own PC

Opera’s browsers can now run open AI models on the user’s computer, through a proposed web standard called WebAI. It is an opt-in early preview, and nothing is downloaded until the user turns it on.

Opera started an early preview of WebAI on October 27. It lets websites and browser extensions use AI models that run on the user’s own computer instead of a company’s servers. It starts in Opera One on Windows and Apple-silicon Macs, with Opera GX and Opera Neon to follow, Opera says.

Moonshot launches Kimi K3.1 Pro, open weights due Nov 4

The Chinese company says its new model nearly matches Claude Opus 5.5 on a coding test. The independent tester Artificial Analysis scores it 55 points, fourth of 215 models, but the weights are not out yet.

Moonshot AI released Kimi K3.1 Pro on October 23, together with a faster version called K3.1 Pro Highspeed. Both work in Kimi’s apps and through its API. Moonshot says it will publish the full model weights by November 4, 2026, so anyone could then download and run the model.

GoTo releases a free AI model for Indonesian and Javanese

The 27-billion-parameter model also handles Sundanese, Balinese and Batak Toba. It is built on Google’s Gemma 3 with AI Singapore, and its weights are free under Google’s Gemma licence.

GoTo, the Indonesian group behind the Gojek and Tokopedia apps, released GoTo-AI-Gemma-3-27B-IT on October 19. The model works in Indonesian, English and four regional languages, and anyone can download it from Hugging Face. GoTo says it scored highest on every regional language in its own tests, but it has published no scores.

DeepSeek releases an open update of its V4.1 Flash model

DeepSeek says the new version scores higher on coding and agent tests, finishes tasks faster and writes shorter answers. Anyone can download it under the MIT licence, and no outside group has tested it yet.

DeepSeek released DeepSeek-V4.1-Flash-1017 on October 17, a “minor update” to the V4.1 Flash model it launched on September 10. The weights are on Hugging Face, and the API price stays the same. In DeepSeek’s own table, the model scores 77.0 on the DeepSWE coding test, up from 74.2.

China proposes export limits on 23 AI technologies

A draft adds 40 technologies in AI, solar power and biomedicine to the list of China’s banned or licensed exports. Offering a model abroad through an API or an app would not count as an export, the commerce ministry says.

China’s commerce and science ministries proposed on October 16 to add 40 technologies to the list of technologies that are banned from export or need a licence. Twenty-three of them are in AI, including large models, AI agents and computing clusters. The public can comment until November 15, and exporters must register these technologies from now on.

Meta releases Muse Spark Open, an 816B open-weight model

Meta calls it the most capable open-weight model from a US company. An independent test gives it 39 points, and in Meta’s own table two Chinese open models still lead on coding.

Meta released the weights of Muse Spark Open on October 15. The model has 816 billion parameters, of which about 42 billion work on each token, and it can read up to 1 million tokens at once. Meta worked with Apple so the model can run on two or four linked Mac Studio computers.

Anthropic says small AI models may soon spread like worms

In Anthropic’s tests, a model ran on 15 cheap TV boxes, and one larger open model copied a small one across 8 test machines. No such AI worm has been seen in the wild yet.

Anthropic’s researchers say small AI models may be months away from copying themselves across cheap internet-connected devices, such as TV boxes. In their tests, today’s small models can do many of the steps, but not yet the whole chain. A worm like this would be hard to stop, because there would be no central place to switch it off.

Mistral previews Large 4, a 1-trillion-parameter model

The French lab says free weights will follow by the end of October. An outside test ranks the preview 64th of 225 models, above the middle of the field but far from the top.

Mistral AI opened a public preview of Mistral Large 4 on October 6. The model has 1 trillion parameters, of which 49 billion work on each token. Mistral says it will release the weights by the end of October, after testing the model with security experts and governments.

Reflection unveils Beam, an open model to rival China’s labs

The US start-up says its 501-billion-parameter model matches a top Chinese model with a quarter to a third of the computing. The weights come later this month, and no outside test exists yet.

Reflection AI announced Beam on October 5, its first open-weight model. The company says Beam scores like Z.ai’s GLM-5.2 on hard reasoning tests while using 3 to 4 times less computing to run. The weights, under the Apache 2.0 licence, are due later this month; for now there is only an early-access list.

Cloudflare releases open Clef models to take on Jev

Clef and the smaller Clef-flash make choices for AI agents and are free to download. Cloudflare says they beat Jev on most of its decision tests and run faster; no outside group has checked yet.

Cloudflare released two open-weight decision models, Clef and Clef-flash, on October 1. They read a situation and a set of typed questions and return decisions, instead of writing text. The weights are on Hugging Face under the Apache 2.0 licence, and both models run on Cloudflare’s Workers AI service, which the company says can stand in for Jev’s.

Amazon releases a free, open rival to the Jev decision model

Strands Decider 2B picks one answer from a fixed list for AI agents and runs on a single graphics card. AWS’s own tests put it near the top of its size class; no outside test has compared it with Jev.

Strands Labs, a team at Amazon Web Services, released Strands Decider 2B on October 1, a small open-source model that makes choices for AI agents. It always picks one option from a fixed list and gives a confidence score, instead of writing text. The weights are on Hugging Face under the Apache 2.0 licence, and AWS says the code, training data and scripts are on GitHub.

Anthropic says China’s GLM-5.3 can build cyber exploits

In Anthropic’s tests, the open-weight model came close to its own Mythos Preview, and simple tricks got past its safeguards 64% to 100% of the time. A US government test had rated it the most cyber-capable open model.

Anthropic said on September 29 that GLM-5.3, an open-weight model from the Chinese company Z.ai, can build working cyber exploits on its own. Its results came close to those of Anthropic’s Claude Mythos Preview. Because anyone can download the model, its safety rules can be removed. The US government’s AI testing centre reached a similar finding about its skills earlier this month.

Image AI maker Black Forest Labs released a robot model

It tops NVIDIA’s RoboLab-120 simulation leaderboard.

Black Forest Labs released FLUX 3 Action, a 7-billion-parameter model that turns camera images and text instructions into robot movements. It leads NVIDIA’s RoboLab-120 leaderboard with 42.9%. The weights are free for non-commercial use, and most speed and real-robot figures come from the company.

Gensyn made a small model whose training can be replayed

The speed cost is large.

Open-1b publishes checkpoints, data, code, and hashes for every training step. Gensyn says audits work across several kinds of hardware, while its reproducible runtime is about five times slower than an optimised system.

Mistral raised €3 billion

Now it must turn European trust into useful AI.

The French AI company reached a value above €21 billion after a new funding round. The money can buy more computing power, but it does not guarantee better models or lasting customers.

MiniCPM5-2B packs long-context AI into a much smaller model

OpenBMB released a 2.5-billion-parameter model for local and on-device use. It also shared much of its training data, but the strongest test results come from the model maker.

Goldman’s CIO says banks can use open AI models

The condition is four layers of safety controls.

Marco Argenti says companies should test models, protect the place where they run, limit agent permissions, and control data access. This is a proposed approach, not a public standard.

Nvidia wants to buy Hugging Face for $12.93 billion

Hugging Face is a major home for open AI models. Nvidia says it will keep the platform open to other chips and clouds, but users and regulators will watch closely.