15 Sources
[1]
Mistral releases weights for its trillion-parameter AI model, targeting cybersecurity
French AI company Mistral has released the weights for its trillion-parameter open source AI model Mistral Large 4 (ML4), allowing businesses and governments to download and deploy the system themselves. Nicknamed "Le Chonk," the model was first introduced earlier this month. The release comes as
[2]
Mistral Says Its AI Model 'Le Chonk' Has Leapfrogged China in the Open-Weight Race
A few weeks ago, French company Mistral released an early version of its new artificial intelligence model, nicknamed Le Chonk. At the time, the company said it was the most advanced open-weight offering outside of China. After further refinement, Mistral now says Le Chonk is the "world's best"
[3]
Mistral's New 'Le Chonk' AI Model Is Big, Open and Now Available for Download - CNET
Mistral Large 4 is a state-of-the-art model for cyberdefense, manufacturing, finance-related tasks and multimodal use. New AI models are coming, and they don't stop coming. On the heels of the recent rollouts of Meta's Muse and Google's Gemini 4 Carbon, Mistral AI is launching its new flagship
[4]
Mistral's Le Chonk is now the top open model for coding - ZDNET
Mistral has released the weights for Large 4, along with some impressive Artificial Analysis scores. * As planned, Mistral has released the weights for its ML4 model. * ML4 sets a new bar for Western open models. * Open models are increasingly competitive with proprietary models. French AI lab
[5]
Mistral open-sources AI model despite warning it could aid cyberattacks
Mistral says later safeguard updates will apply to its managed API and future releases, not to the model copies organizations have already downloaded and deployed. endif; ?> A trillion-parameter AI model that its developer says could help attackers carry out real cyberattacks is now free to
[6]
Mistral AI publicly releases Large 4, targets cybersecurity users
Oct 29 (Reuters) - French AI startup Mistral released its Large 4 model to the public on Thursday, saying it was designed to help organisations identify and fix cyber-security vulnerabilities. Unlike leading AI systems such as Google's (GOOGL.O), opens new tab Gemini, which are mainly accessed
[7]
Mistral releases open weights for its 1 trillion-parameter Large 4 model
Three more weeks of reinforcement learning took the model from 28% to 41% on Terminal-Bench 4.0, and the weights are now on Hugging Face. Mistral has released the weights of Mistral Large 4, the 1 trillion-parameter model it previewed three weeks ago, and the version now on Hugging Face is
[8]
Mistral is taking on U.S. and Chinese AI giants with its biggest model yet
For the past few years, Mistral has tried to carve itself a name as Europe's answer to Silicon Valley's AI giants, moving from smaller, open models to heavyweight systems like Mistral Large 3 in late 2025. After raising €3 billion in September 2026, the French start-up has doubled down on its
[9]
Mistral Large 4 arrives as the top open model outside China with improved coding performance and 63% lower cost-per-task
Mistral has released the downloadable open weights for its trillion-parameter Large 4 model under a custom license, giving enterprises a new option for running and customizing a near-frontier AI model themselves. Independent tests now show that its biggest advances over the preview involve
[10]
Mistral says its new model, Large 4, is the best open model for agentic coding and cyber defense | Fortune
French AI lab Mistral has released it's newest model, Mistral Large 4, to the public -- claiming it has chased its Chinese rivals to the frontier of open-weight AI. Unlike most leading AI models, open-weight models are made freely available for anyone to download, run and modify. It is a class of
[11]
Mistral releases latest AI model for download
The French AI company says its Large 4 model offers advanced coding and cyber-defence capabilities and can be run on an organisation's own computers and servers. Mistral has released the weights for its latest large flagship AI model, Mistral Large 4 (ML4), nicknamed "Le Chonk", claiming
[12]
Mistral releases open-weights for Mistral Large 4, providing sovereign control for the company's 'Le Chonk' AI model
Mistral AI SAS today announced the availability of the open weights of its frontier model Mistral Large 4, nicknamed "Le Chonk," the company's most capable large language model to date, initially released in preview three weeks ago. With 1 trillion parameters and 52 billion active parameters, ML4
[13]
Mistral AI publicly releases Large 4, targets cybersecurity users
French AI startup Mistral released its Large 4 model to the public on Thursday, saying it was designed to help organisations identify and fix cyber-security vulnerabilities. Unlike leading AI systems such as Google's Gemini, which are mainly accessed through online services, the model can be
[14]
Mistral Touts Cyber Capabilities of Its 'Chonk' AI Model | PYMNTS.com
The model builds on a version released in public preview earlier this month, the French startup said in the announcement. "The response to the preview has exceeded our expectations, and we're grateful to our developer community, early-access customers, and partners for their support, collaboration
[15]
Mistral AI publicly releases Large 4, targets cybersecurity users
Oct 29 (Reuters) - French AI startup Mistral released its Large 4 model to the public on Thursday, saying it was designed to help organisations identify and fix cyber-security vulnerabilities. Unlike leading AI systems such as Google's Gemini, which are mainly accessed through online services, the
Share
Copy Link
French AI company Mistral AI has released the weights for its trillion-parameter open-weight AI model, Mistral Large 4, nicknamed Le Chonk. The model targets cybersecurity and agentic coding tasks, claiming to outperform Chinese competitors while using 47% fewer tokens. However, Mistral acknowledges the model could aid real cyberattacks despite built-in safeguards.
French AI company Mistral AI has released the weights for its trillion-parameter open-weight AI model, Mistral Large 4 (ML4), nicknamed Le Chonk
1
. The release allows businesses and governments to download, deploy, and own the model independently, marking a significant milestone in the competition between open-source and closed AI providers. The model was initially introduced in public preview earlier this month and has undergone weeks of additional reinforcement learning and fine-tuning based on feedback from developers, cybersecurity experts, and state officials3
. By making the weights available on Hugging Face and via its API1
, Mistral AI positions Le Chonk as a tool for empowerment rather than dominance, enabling organizations to maintain control over their data and systems without relying on external AI companies.
Source: PYMNTS
Mistral AI asserts that Le Chonk has achieved state-of-the-art performance among open-weight AI models, particularly excelling in agentic coding tasks and sovereign cyber defense capabilities. The model scored 41% on Terminal-Bench 4.0, surpassing the Chinese competitor GLM-5.3 by one point to claim the top spot for AI model for coding agents
4
. This represents a significant jump from the preview version's 28% score just weeks earlier. Mistral AI emphasizes that Le Chonk rivals top open models while using 47% fewer tokens per task on average compared to GLM-5.31
, making it significantly more economical. The company also claims Le Chonk outranks all other open models on Artificial Analysis' Coding Agent Index1
, demonstrating its capabilities in planning, research, and handling complex enterprise tasks that would typically take human experts hours or days to complete3
.
Source: VentureBeat
Mistral Large 4 was specifically designed to excel at defensive cybersecurity tasks, including log inspection, code auditing, and checking for newly published exploits or vulnerabilities
3
. The model scored just above DeepSeek V4.1 Flash on the CyberGym benchmark for defensive cybersecurity, which measures how well models find real memory-safety vulnerabilities in C/C++ open-source projects4
. Pierre Stock, Vice President of Science at Mistral AI, explained that the open-weight nature of the model ensures organizations can access frontier intelligence to defend against cyberattacks without risk of access being cut off1
. The model enables real-time threat detection and response to major security incidents, with the ability to be deployed on-premises within an organization's own ecosystem1
. Stock emphasized the importance of providing this capability to everyone as soon as possible to strengthen the ecosystem against malicious actors who might leverage their own technology to attack1
.Despite its defensive focus, Mistral AI acknowledges that Le Chonk could potentially aid real cyberattacks, highlighting the dual-use nature of AI technology. The company stated that while monitoring cybersecurity-related use during the preview phase, they observed the model refusing to comply with black-box attacks against live targets, but noted a potential for uplift to conduct real attacks
5
. Mistral AI claims the model refused 96.6% of malicious prompts in cyber-related portions of benchmarks including JailbreakBench, StrongREJECT, and AgentHarm5
. However, subsequent safeguard updates will only apply to Mistral's managed API and future releases, not to model copies that organizations have already downloaded and deployed5
. Security experts note that freely available models can be downloaded and customized by threat actors, presenting clear risks alongside productive uses2
.Related Stories
The release of Le Chonk represents what Stock calls a pivotal moment in the global AI leadership race, particularly in the context of geopolitical implications between Western and Chinese AI development
2
. Mistral AI argues that businesses and governments need access to advanced open-weight models that cannot be switched off remotely, unlike proprietary models from OpenAI, Anthropic, and Google2
. Stock emphasized that AI will become like electricity, and organizations shouldn't live in a world where access can be cut off from one day to another2
. The model was trained from scratch on 4,000 Nvidia Grace Blackwell GPUs over two months in Mistral's own data centers in Europe4
, demonstrating significant compute efficiency compared to larger training runs. Mistral AI also emphasizes data sovereignty concerns, suggesting that Chinese models could potentially behave maliciously or serve as threat vectors2
.Source: The Next Web
Beyond cybersecurity and coding, Mistral Large 4 was built to handle multimodal tasks including processing both text and images. The model demonstrates advanced capabilities in geospatial analysis, outperforming competitors in aerial and satellite imagery analysis with the capacity to identify visual patterns and objects on maps, such as forest fires
3
. The model also excels at grounding capabilities, which help eliminate hallucinations by anchoring outputs in external resources like documents and databases rather than relying solely on internal training materials3
. Mistral AI trained the model on more than 160 languages, including every official language of the European Union, as well as Latin and non-Latin languages3
. The model is available through multiple channels including Mistral Studio, the Mistral API, OpenRouter, and Nvidia-hosted endpoints3
. Additionally, Mistral is handling a specialized cybersecurity variant differently, providing preview access to design partners such as Orange Cyberdefense for penetration testing and vulnerability assessment, subject to rigorous vetting and contractual safeguards5
. Under the EU AI Act, models trained with more than 10^25 floating-point operations are presumed to carry systemic risk, placing Le Chonk under regulatory scrutiny5
.Summarized by
Navi
06 Oct 2026•Technology

02 Dec 2025•Technology

18 Mar 2025•Technology
