Search “OpenAI”

Mistral Open-Sources Mistral Large 4 Weights, Leading GLM-5.3 by 1 Point on Coding Benchmark

Mistral released the weights of its frontier open-source model Mistral Large 4 (ML4) on Hugging Face on Thursday; the model had only been available in preview three weeks earlier. Mistral calls ML4 "the world's best open-source agentic coding model," scoring 41% on Terminal-Bench 4.0, slightly above GLM-5.3's 40%, and says it uses an average of 47% fewer tokens per task than GLM-5.3; it scored 79% on the CyberGym E2E defensive cybersecurity test, compared with 29% for GLM-5.3. Mistral Vice President of Science Pierre Stock called ML4 merely a "stepping stone" and confirmed that smaller versions will be released in the coming months, one of which can run on a laptop.

Mistral says its trillion-parameter open-source model Le Chonk surpasses Chinese models in coding and cybersecurity

Mistral has released the official version of Le Chonk, a trillion-parameter open-weight model, calling it the world's best open-weight model in automated coding and cybersecurity, and making it freely available. Pierre Stock, Mistral's VP of Science, said the model was trained on increasingly difficult cybersecurity and coding problems, enabling it to match or surpass models from Chinese labs such as DeepSeek and Z.ai. Stock also stressed that the West needs its own advanced open-weight models, because closed models can be shut down remotely, whereas open weights cannot have access cut off once released.

Google DeepMind releases Gemini 4 Carbon, topping Artificial…

Google DeepMind has released Gemini 4 Carbon, which ranks first on the Artificial Analysis Intelligence Index v5 with a score of 59 and leads in 4 of 9 evaluations including HLE-Diamond and TB-Science. The model focuses on cybersecurity and long-horizon tasks, claiming 88.2% on SEC-bench Pro, 54.8% on ExploitGym, and 71.0% on CWE-bench, and is initially available only to trusted cybersecurity personnel through the Fairwind Program. In terms of specifications, it supports a 1M token context and 1M maximum output, with input covering text, images, video, and speech, and output in text; Artificial Analysis lists a standard price of $4/$20 per million input/output tokens, plus a 50% launch discount.

Google Releases Gemini 4 Carbon, Surpassing Claude Opus 5.5 on Multiple Benchmarks

Google has released its new flagship model Gemini 4 Carbon, which tops the Artificial Analysis Intelligence Index 5.0 with a score of 59, surpassing Claude Opus 5.5 and GPT-6 Astra. It scored 65.2% on Terminal-Bench 4.0, 96.2% on GraphWalks for 256k to 1M context, and 88.2% on SEC-bench Pro vulnerability discovery. Initial pricing is $2 per million tokens for input and $10 per million tokens for output, with cached input as low as $0.2; because of its vulnerability discovery and exploitation capabilities, it will first be made available to select cybersecurity defenders.

NVIDIA Releases Nemotron 3.5 Transcribe: An Open-Source 30B MoE Speech Model…

NVIDIA has released Nemotron 3.5 Transcribe, an open-weight large speech model that supports 40 languages and up to 1 hour of audio per pass, and can select output formats such as verbatim, written, speaker-labeled, contextual, and translation via text instructions. The model has approximately 34B total parameters and about 4.3B activated parameters, adopts a hybrid Mamba-Transformer MoE architecture, can run in BF16 on a single H100 80GB, and the NVFP4 version at about 21GB can run on an RTX 5090. NVIDIA self-reports an average WER of 4.71% across 9 public English Open ASR test sets, the best result among open-weight models, though commercial APIs still lead; its shortcomings are that it only supports offline operation, and the Chinese far-field meeting cpCER is 34.95%.

Block Releases Dessa, a Fully Duplex Speech Model Trained Purely on Speech

Block's AI research lab has released Dessa, a full-duplex speech model, opening a waitlist in the form of a research preview. It is trained only on human voice conversation audio, without using text or pretrained models. The model can run locally on a MacBook Pro M4 Max or higher configuration, faster than real time; training used one NVIDIA GB200 NVL72 (72 GPUs). No benchmark scores, parameter count, weights, paper, or API have been published so far.

Google DeepMind releases the Genie 4 world model, which can generate sound effects and converse with characters in the scene

Google DeepMind has released Genie 4, a new-generation world model that can turn prompts into explorable, realistic scenes in real time, and for the first time the Genie model itself generates background sounds and music, while characters in the scene can also hold real-time conversations. Users move via the keyboard and use text prompts to modify content in the scene instantly, with the scene remaining coherent across multiple inputs. In the demo, a simulated Apptronik Apollo robot plays checkers, with each move planned by Gemini 4 Argon, pointing to Genie's future use in robot training.

Mistral Releases Weights for Trillion-Parameter Open Model Mistral Large 4, Focused on Cybersecurity

Mistral has released the weights for its trillion-parameter open-source model Mistral Large 4 (ML4, nicknamed Le Chonk), which enterprises and government agencies can download and deploy themselves; the weights are already available on Hugging Face and via its API. Mistral says the model's performance is comparable to that of the open-source model GLM-5.3, but it consumes 47% fewer tokens per task, and it ranks first among open-source models on Artificial Analysis's Coding Agent Index. Pierre Stock, Mistral's Vice President of Science, said that open weights mean organizations can deploy locally in their own environments to identify, coordinate, and respond to major security incidents in real time.

Google releases Gemini 4 Carbon, tops the Artificial Analysis leaderboard with 59 points

Google released Gemini 4 Carbon, which scored 59 points on the Artificial Analysis Intelligence Index v5.0 leaderboard, surpassing Claude Opus 5.5's 58 points and GPT-6 Astra's 57 points. The model achieved 88.2%, 54.8%, and 71.0% on SEC-bench Pro, ExploitGym, and CWE-bench, respectively, and Google says it can handle engineering problems in large codebases continuously for several days. The model this week is first being made available to some cybersecurity defenders, then expanded to trusted partners through the Fairwind Program, with no timeline yet for general users; initial pricing is $2 per million input tokens and $10 per million output tokens.

Block launches full-duplex speech model Dessa, supports local operation

Block has launched the full-duplex speech model Dessa, which can listen to the user's speech while speaking, and can simulate emotional expressions such as laughter, pauses, and choking up. The company says it is trained entirely on human speech data, using only a single rack of NVIDIA GPUs for training, and can run locally on MacBooks with M4 Max and higher configurations. At this stage Dessa is not open for public trial; Block is sending testing invitations to security researchers, focusing on risk scenarios in which speech agents deceive humans.

Block releases full-duplex voice model Dessa, which can run locally on consumer-grade GPUs

Block's research lab has released the voice model Dessa, which uses a full-duplex architecture, can listen and speak at the same time, and can run locally on a MacBook M4 Max or a more powerful consumer-grade computer. The model was trained entirely on human speech and a single NVIDIA GPU, and the company says expressions such as emotion, pauses, and laughter are built into the model rather than driven by a text pipeline. Dessa is not currently publicly available, and Block is recruiting security researchers to test the risk that it could be used to deceive conversation partners.

Moonshot AI releases Kimi K3.1 Pro: 2.8T total parameters, 104B activated parameters, weights scheduled to be opened on November 4

Moonshot AI releases Kimi K3.1 Pro, a multimodal MoE model with 2.8T total parameters, 104B activated parameters, and a 1 million token context window. The Standard and Highspeed versions are now available on the Kimi website, App, Kimi Work, Kimi Code and API. The company says Highspeed offers 2–3x faster output at twice the price of the Standard version. Artificial Analysis ranks it fourth on the Intelligence Index with a score of 55, behind Opus 5.5's 58; on Arena's Code Arena WebDev it ranks 8th with 1692 points, up nine places from K3. Standard pricing is $2 per million tokens for input, $10 for output, and $0.2 for cached input, one-third lower than K3; the weights have not yet been released, and the company promises to open them on November 4.

Moonshot AI releases Kimi K3.1 Pro: a 3T-class open-weight model with a 1 million token context

Moonshot AI releases Kimi K3.1 Pro, the first model in the K3.1 series and its strongest 3T-class open-weight model to date, with a 1 million token context window, alongside Kimi K3.1 Pro Highspeed, which is 2-3 times faster. The model is now available on Kimi.com, Kimi Work, Kimi Code and the Kimi API, with full model weights to be released on November 4. On Terminal-Bench 4.0, K3.1 Pro scores 65.7%, higher than GPT-6 Astra's 58.2% and slightly below Claude Opus 5.5's 66.4%; on the internal benchmark Kimi Code Bench 3.0 Long-horizon it scores 61.2%, higher than K3's 55.95%.

Hugging Face Releases Open-Source DNA Model Carbon-X, Runs Locally on a Laptop

Hugging Face on Thursday released the open-source DNA understanding model Carbon-X, which can run locally on a laptop, letting users ask in natural language how DNA changes affect gene activity and health. The model was developed in collaboration with researchers at Zhongguancun Academy and the University of Naples Federico II; it extends the vocabulary of an existing language model to read DNA in six-letter units and was further trained on 300 billion tokens of DNA sequences, English text, and samples pairing DNA with biological descriptions. The model has 14B parameters and can process sequences of about 50,000 base pairs at a time; it cannot upload a complete human genome. The developers also note that the prediction accuracy has not yet been validated through wet-lab experiments.

Hugging Face Releases Open-Source Model Carbon-X, Which Can Analyze and Generate DNA Sequences Using Natural Language

Hugging Face has released the open-source model Carbon-X, combining DNA analysis and generation with natural language so that researchers can ask questions directly and receive predictions and explanations. The model is based on Qwen3-14B and was trained on genomic sequences and natural language from humans, other mammals, plants, and fungi, with viruses explicitly excluded from the training data. Researchers can run it locally, avoiding the sharing of sensitive patient data with external AI companies, and can also generate new DNA sequences to assist gene therapy research.

Recursive releases laptop-class coding model Flint, says it was built mainly with AI involvement

AI startup Recursive has released its first model, Flint, a coding model that runs offline on a laptop and is freely downloadable on Hugging Face under the Apache 2.0 license; the company says its development process was mainly carried out with the involvement of its in-house Eureka Machine. Flint is an MoE model that activates only 2.5B of its 40B parameters at a time, generates at about 131 tokens/sec on a MacBook Pro with an M4 Pro, and outperforms other models under 50B parameters on SWE-Bench Verified. Recursive says Flint requires more guidance than large paid coding assistants and recommends breaking large tasks into small steps.

Inception Releases Diffusion Language Model Mercury 3, Claiming Speeds Four to Eight Times Faster Than GPT-6 Luna

Inception has released the diffusion language model Mercury 3, claiming it wins 8 of 12 benchmarks and runs four to eight times faster than GPT-6 Luna and Claude Haiku 5.5. In Inception's own testing, Mercury 3 scored 76.84 on IFBench, 15 points higher than GPT-6 Luna; it scored 51.08 on SWE-Bench Atlas, versus 33.33 and 32.26 for its competitors. Claude Haiku 5.5 led on the DeepSWE long-code test and long-text retrieval. The model generates 1,068 tokens per second on Nvidia GPUs, is priced at $0.20 per million input tokens and $0.70 per million output tokens, with a 50% discount during the launch period, and is now available on the Inception platform as well as OpenRouter, Baseten, and Vercel; there are currently no independent tests.

English translation is being prepared.

Memory Tensor Open-Sources MemT 1.0 Series Memory-Native Models

Memory Tensor has open-sourced and released the MemT 1.0 series of memory-native models, including MemT-1.0-Nano with 36B parameters and 3B activated (based on Qwen3.6-35B-A3B) and MemT-1.0-Flash with 288B parameters and 17B activated (based on DeepSeek-V4-Flash), adding a memory computation pathway parallel to attention in selected Transformer modules. The team built a 34-billion-token memory training corpus, of which 20 billion tokens were used for continued pre-training and 14 billion for post-training. In a comprehensive evaluation across 11 memory benchmarks, Flash and Nano achieved average scores of 50.6 and 47.0, respectively, ranking in the top two among the parametric memory methods evaluated; on 500 SWE-bench Verified tasks, the proportion of tasks that Nano succeeded on all three attempts rose from 54.00% for the no-experience baseline to 62.80%.

MiniMax Releases M3.1 Flash Model, ALE Score of 29.6% Ranks First Among Domestic Models

MiniMax officially released M3.1 Flash, formerly known as Space Bunny (Yutu), which previously topped the charts anonymously on OpenRouter. The model scores 29.6% on Agents' Last Exam, ranking first among domestic models, and scores 43 on the AA leaderboard; it achieves 73.7% on DeepSWE v1.1, 82.2 points on CyberGym, 76.2 points on BankerToolBench, tying for first with Claude Opus 5, and 61.7 points on OSWorld V2. The model adopts a sparse attention architecture, with output speeds of up to 200+ TPS and a median of 170+ TPS, input at 1 yuan per million tokens, output at 4 yuan, and cache reads at 0.1 yuan, with comprehensive usage costs reduced by 70% compared to the previous generation M3.