Mistral opens the weights of the trillion-parameter model Le Chonk

1. 17:58
   WebAI Threat Watch (aithreat.io)· Media coverage

   Mistral open-sources trillion-parameter model, self-reports it may aid cyberattacks - Artificial Intelligence(opens the original post in a new tab)
2. Time not given
   zhiding.cnSelected

   Mistral open-sources trillion-parameter model Large 4, says it may assist real cyberattacks

   Mistral AI released the open weights of its trillion-parameter model Mistral Large 4, and stated that the model has the potential to assist in carrying out real cyberattacks, while saying that traffic monitoring during the preview stage showed it refused black-box attacks against real targets. The company said subsequent guardrail updates will only apply to the hosted API and future released models, and will not automatically update already-downloaded checkpoints. On the cybersecurity-related portions of JailbreakBench, StrongREJECT and AgentHarm, Large 4 refused 96.6% of malicious prompts, which Mistral says is the highest refusal rate it has measured among open models; a Gartner analyst cautioned that the figure only reflects performance in its test scenarios. A security variant optimized for penetration testing, vulnerability assessment and red-team exercises is available in preview only to design partners who sign contractual assurance terms.

1. 22:40
   X@ApiSerpent· Discussion

   Mistral released Le Chonk yesterday. 1 trillion parameters. Open weights. On Hugging Face. Top open-weight coding scor…(opens the original post in a new tab)
2. 22:18
   X@TheCharismaCat1· Discussion

   Mistral says it has released the weights of Mistral Large 4, a 1-trillion-parameter model under a modified MIT licence.…(opens the original post in a new tab)
3. 15:20
   X@polderbase· Discussion

   Mistral released the weights for Mistral Large 4: a frontier-grade model you can run and own. The gap between closed AP…(opens the original post in a new tab)
4. 15:01
   X@AxSaucedo· Discussion

   Mistral Large 4 "Le Chonk" is now open weights 🇪🇺🚀🇫🇷 This is Mistral's 1 trillion parameter multimodal MoE, with three…(opens the original post in a new tab)
5. 14:30
   X@hfcxgexf· Discussion

   The biggest structural news in the AI world this week may not be which model a big company released, but an open-source model: Le Chonk, 1T parameters, fully open source, leading on coding benchmarks. Self-hosting players finally have a big one that can compete—no longer having to rent APIs and defer to big companies. My judgment: every bit stronger an open-source model gets, the big companies' pricing power…(opens the original post in a new tab)
6. 14:00
   X@dagcannural· Discussion

   1 trilyon parametrelik bir model artık indirilebilir 🏔 Mistral, Large 4'ün ağırlıklarını Hugging Face'te yayınladı. A…(opens the original post in a new tab)
7. 04:17
   X@AlexVoss88· Discussion

   298 gigabytes. That is the smallest download for Mistral's "open" trillion-parameter model, and the fine print is bigge…(opens the original post in a new tab)

1. Time not given
   Mistral AI

   Mistral Large 4 (Le Chonk) brings intelligence, Mistral Studio puts it to work

   Mistral launches Mistral Large 4 (Le Chonk) and expands the Mistral Studio enterprise platform. The new model excels at coding, defensive cybersecurity, and reasoning over complex documents and business tasks; the platform adds hosted agents and applications, workflows, enterprise context, observability, and evaluation features. The platform supports cloud, VPC, or on-premises deployment while retaining full control, and ABANCA, BNP Paribas, Capgemini and others are already using it in production.

2. 03:47
   X@_moelevin· Discussion

   Mistral opened the weights on its best bug-hunting model and called it Le Chonk. Moonshot won't release anything like t…(opens the original post in a new tab)

1. 23:12
   X@Connorclaw· Discussion

   Mistral arrives at coding club wielding her baguette sword. 🥖⚔ Mistral Large 4 “Le Chonk” now has open weights, focus…(opens the original post in a new tab)
2. 20:18
   X@checkZaranz4u· Discussion

   Mistral's official account just showcased Large 4: 133 agents divide the work on the latest Django CVE to explore the code and reproduce the vulnerability, then Le Chonk directly applies the patch, the whole process being triage → repro → patch. Put plainly, this security research work is also about to open…(opens the original post in a new tab)
3. 14:30
   X@petediano· Discussion

   Mistral just released the open weights for its 1-trillion-parameter model “Le Chonk”. It leads open models on coding a…(opens the original post in a new tab)

1. 20:30
   X@firethewalla· Discussion

   Le Chonk is out of the cage. Mistral just released the open weights for Large 4 — 1 trillion parameters, natively mult…(opens the original post in a new tab)
2. 16:25
   X@420dotmeme· Discussion

   mistral dropped a 1-trillion-parameter model called le chonk. 49B awake at a time, 1M context, open weights. still just…(opens the original post in a new tab)
3. 16:22
   X@ChinaAIDaily· Discussion

   ③ What happened with Mistral's open-source trillion-parameter large model Le Chonk: On October 29, France's Mistral opened the weights of Mistral Large 4 (Le Chonk): trillion parameters, native multimodal, about 52 billion active, trained in-house in Europe. Highlights: called…(opens the original post in a new tab)
4. 09:18
   X@zhoushengjin21· Discussion

   Mistral open-sources the 1T-parameter Le Chonk: Europe's open-source flagship has finally landed, with a dual push in agentic coding and cyber defense. Official blog: https://mistral.ai/news/mistral-large-4-oss/ Simply put, this is currently the agentic coding… among open-weight models(opens the original post in a new tab)
5. 06:54
   X@yeshcorp· Discussion

   Mistral released the weights for a new Le Chonk model, built for agentic coding and cybersecurity. Open weights means…(opens the original post in a new tab)
6. 06:50
   X@bymachine_news· Discussion

   Mistral Releases Open Weights for 1T-Parameter Le Chonk Model https://bymachine.news/mistral-large-4-le-chonk-open-weig…(opens the original post in a new tab)
7. 06:49
   X@Mintscope1· Discussion

   Mistral released the open weights of Mistral Large 4 (Oct 29): ~1T parameters, 49B active, 1M-token context. Free to se…(opens the original post in a new tab)
8. 06:30
   X@xsunset· Discussion

   Mistral released the weights of the trillion-parameter Large 4 directly, freely downloadable on Hugging Face, nicknamed Le Chonk. Most striking is that the official side itself admitted first: this model has the potential to improve real attack capabilities, observed during the preview period. Even more striking, copies that have been downloaded, the guardrails afterward…(opens the original post in a new tab)
9. 06:10
   X@mdp_sec· Discussion

   mistral just put out the le chonk weights and aimed it straight at defenders the preview cost $1.31 a solve on my 100 b…(opens the original post in a new tab)
10. 05:41
    X@MadeInBot· Discussion

    Le Chonk is real, it has 1 trillion parameters, and you can download it. 🐘 Hugging Face sat down with Mistral's team o…(opens the original post in a new tab)
11. 05:28
    X@marctechireai· Discussion

    Foundation Models: @MistralAI put the weights for Large 4 on Hugging Face. 1T params, 52B active, and pitched at agenti…(opens the original post in a new tab)
12. 04:30
    X@TeksEdge· Discussion

    Mistral Large 4, aka Le Chonk, is now live on llama.cpp (a 1T-parameter model). And the “4-bit” GGUF is ~619GB. 👀 Le…(opens the original post in a new tab)
13. 04:27
    X@AI_News_Digest· Discussion

    Mistral releases open weights for its 1T-parameter Large 4 model nicknamed Le Chonk with 1M context. #AI #AINews https:…(opens the original post in a new tab)
14. 03:45
    X@wikiwayne· Discussion

    Mistral just opened the weights for Large 4, "Le Chonk": 1T total params, 49B active per token, 1M context. Local real…(opens the original post in a new tab)
15. 03:44
    X@Gas4rmda4· Discussion

    Mistral 🇫🇷 Just dropped the weights for their 1 TRILLION parameter model "Le Chonk" on Hugging Face 😳 Built for coding…(opens the original post in a new tab)
16. 03:41
    X@kevintame· Discussion

    Mistral released the weights for Large 4 ("Le Chonk"), a trillion-parameter model you can run on-prem. They're pitchin…(opens the original post in a new tab)
17. 03:16
    X@SovGlobalMag· Discussion

    Mistral, the Paris AI lab, has released the weights of Large 4, nicknamed Le Chonk. It is now the highest-scoring open…(opens the original post in a new tab)
18. 02:50
    X@phgodoycosta· Discussion

    A Mistral acabou de abriu os pesos do Large 4. 1 🤩 1 trilhão de parâmetros com peso aberto pra rodar onde você quiser.…(opens the original post in a new tab)
19. 02:37
    X@BMTestsAI· Discussion

    Google locked its best cyber model away. Mistral just gave its away. Mistral Large 4's weights are out: 1 trillion par…(opens the original post in a new tab)
20. 02:32
    X@AiPoJuGe· DiscussionIn Chinese, as written

    Mistral 开源了 1 万亿参数的 Le Chonk 欧洲也想在 frontier 模型里抢一席之地 闭源 vs 开源，2026 这仗才刚开打。你押哪边？(opens the original post in a new tab)
21. 01:56
    X@_MiladB· Discussion

    The most underrated feature in AI right now? Being able to leave. Mistral just released the weights for its 1-trillion…(opens the original post in a new tab)
22. 01:55
    X@compuchris· Discussion

    Mistral Releases 1-Trillion-Parameter AI Model for Cybersecurity https://www.techrepublic.com/article/news-mistral-larg…(opens the original post in a new tab)
23. 01:51
    X@kaddisdeployed· Discussion

    MISTRAL JUST DROPPED THE WEIGHTS FOR LARGE 4 (LE CHONK) 🔥 we're talking about a nearly 1 trillion parameter multimodal…(opens the original post in a new tab)
24. 01:44
    X@saikrishna__777· Discussion

    Mistral just released the open weights for its 1-trillion parameter Large 4 model (Le Chonk)! 🚀 It's a powerful multim…(opens the original post in a new tab)
25. 01:40
    X@bijaypandey· Discussion

    Mistral just dropped the weights for Large 4, nicknamed Le Chonk: a trillion parameters, 52B active, downloadable from…(opens the original post in a new tab)
26. 01:18
    X@DoDataThings· Discussion

    Best non-Chinese open model: 26. Mistral Large 4's weights are out (1.05T total, 52B active). It narrowly takes #1 out…(opens the original post in a new tab)
27. 01:08
    X@amjustadam· Discussion

    Mistral Releases Open Weights for Le Chonk, Its One Trillion Parameter Mistral Large 4 Model https://x.com/MistralAI/st…(opens the original post in a new tab)
28. 01:02
    X@mrru5s3ll· Discussion

    Mistral Large 4 (Le Chonk) weights are now out. 1T total parameters, roughly 50B active, 1M token context window. The b…(opens the original post in a new tab)
29. 01:00
    X@themuqsit· Discussion

    Mistral released the weights of Le Chonk, a new open model for agentic coding and cybersecurity. Mistral says it ranks…(opens the original post in a new tab)
30. 00:59
    X@SarathNair82· Discussion

    8/8 Sources (every number above is from these, plus OpenRouter's model listing): Mistral launch post https://mistral.a…(opens the original post in a new tab)
31. 00:52
    X@SarathNair82· Discussion

    2/8 What it is: Mistral Large 4, nicknamed "Le Chonk". - ~1T total params, A52B active (MoE, 128 routed experts, 4 per…(opens the original post in a new tab)
32. 00:45
    X@techamericaofcl· Discussion

    Mistral AI has released its trillion-parameter Large 4 model with downloadable weights, giving enterprises greater cont…(opens the original post in a new tab)
33. 00:23
    X@AliLenjani· Discussion

    Releasing open weights for a 1-trillion parameter model is a massive push for coding agents. If three weeks of extra RL…(opens the original post in a new tab)
34. 00:01
    X@TraTTow_br· Discussion

    Mistral open-sourced the weights for Mistral Large 4, a trillion-parameter security model. Run and tune it in your own…(opens the original post in a new tab)
35. 00:01
    X@BareStack· Discussion

    Mistral released the weights for its trillion parameter model. They call it Le Chonk. Download it. Run it on your own…(opens the original post in a new tab)
36. 00:01
    X@nstlbcn· Discussion

    Mistral allibera els pesos oberts del model Large 4 d'1 bilió de paràmetres. Accelerarà l'adopció de models oberts? Fon…(opens the original post in a new tab)
37. 00:00
    X@SamElonJr· Discussion

    Mistral released open weights for a cybersecurity AI called Le Chonk. Somewhere a CISO has to explain to the board tha…(opens the original post in a new tab)
38. 00:00
    X@zotpaper· Discussion

    Mistral releases Large 4 weights. Mistral says it released the weights of Mistral Large 4, also called Le Chonk, and c…(opens the original post in a new tab)
39. 00:00
    X@EcoDa172493· Discussion

    Open weights mean freedom—and a new chore: guarding every copy. Mistral’s 1T-parameter Large 4 targets coding and cyber…(opens the original post in a new tab)

1. 23:49
   X@Lab_Sense· Discussion

   6/ Le Chonk sort de sa boîte 🇪🇺 Mistral ouvre les poids de Large 4 : 1 000 Md de paramètres, entraîné en Europe, au top…(opens the original post in a new tab)
2. 23:45
   X@bretts77011· Discussion

   Mistral released the open weights for Large 4 ('Le Chonk'): a ~1T-parameter MoE model aimed at agentic coding and defen…(opens the original post in a new tab)
3. 23:41
   X@yibie· Discussion

   On October 29, Mistral released the weights of its flagship model Large 4 (nicknamed Le Chonk) on Hugging Face. This is a trillion-parameter open-weight model, with coding and defensive cybersecurity as its main selling points. Before deciding whether to download it and run it in your own data center, consider four accounts first. The…(opens the original post in a new tab)
4. 23:33
   X@Tajudeen· Discussion

   … @mistralai has released the weights for its large 4 model, nicknamed “le chonk.” and the name makes sense when you l…(opens the original post in a new tab)
5. 23:31
   X@QuidVanNovi· Discussion

   Mistral just dropped the weights for Le Chonk, a 1 trillion parameter model, and told everyone to download it. It has…(opens the original post in a new tab)
6. 23:22
   X@Z3nlyte· Discussion

   mistral put a trillion parameters on hugging face. large 4 open weights: 52B active, 1M context, modified MIT with a mo…(opens the original post in a new tab)
7. 23:15
   X@JaymianLee_· Discussion

   $ ai_news --major Mistral released the weights of Large 4 ("Le Chonk"). 1 trillion parameters. Open weights. Mistral…(opens the original post in a new tab)
8. 23:11
   X@orfa1017· Discussion

   Mistral directly released the weights of the trillion-parameter Large 4, codenamed Le Chonk—what a name. 1 trillion parameters, a MoE activating about 49B per token, 1 million context, focused on coding agents and cybersecurity. Enterprises and governments can deploy it themselves, but wanting to take it and…(opens the original post in a new tab)
9. 23:04
   X@ycnewsdigest· Discussion

   Mistral (@MistralAI) just dropped the 1‑trillion‑parameter Mistral Large 4 (Le Chonk) as open‑weight, claiming top open…(opens the original post in a new tab)
10. 23:02
    X@dylan2045ad· Discussion

    Podcast📣Mistral Al released the open weights of its trillion-parameter Large 4 model, nicknamed Le Chonk, on October 29…(opens the original post in a new tab)
11. 22:44
    X@saikrishna__777· Discussion

    Mistral just released the open weights of its flagship Large 4 model, nicknamed Le Chonk! 🐱 It's a ~1 trillion paramet…(opens the original post in a new tab)
12. 22:14
    X@li_ke_nan· Discussion

    Mistral open-sources all weights of Mistral Large 4 (Le Chonk), 1 trillion parameters 1T parameters fully open weights, unimaginable two years ago The key is turning "sovereign AI" into a business: self-hostable, customizable, trained in Europe For teams doing agent programming and security defense, this is a closed-source AP...(opens the original post in a new tab)
13. 22:13
    X@ForkLog· Discussion

    French company Mistral AI has released the weights for its flagship artificial intelligence model, Large 4 (Le Chonk),…(opens the original post in a new tab)
14. 22:12
    X@AsWill11· Discussion

    mistral just open-weighted le chonk (mistral large 4) — 1T moe (~49–52B active), native multimodal, 1M context. trained…(opens the original post in a new tab)
15. 22:06
    X@painandtears3· Discussion

    Europe just dropped a trillion-parameter open model. @MistralAI released weights for Large 4, "Le Chonk": 1.05T total,…(opens the original post in a new tab)
16. 22:01
    X@TraffAlex· Discussion

    🤖 AI NEWS ROUNDUP — October 30, 2026 1⃣ MISTRAL OPENS THE WEIGHTS ON ITS TRILLION PARAMETER FLAGSHIP Mistral release…(opens the original post in a new tab)
17. 22:00
    X@overpr0mptedd· Discussion

    Mistral just put the weights of a 1-trillion-parameter model on Hugging Face. Mistral Large 4 ("Le Chonk"): - 1T param…(opens the original post in a new tab)
18. 21:52
    X@mark_k· Discussion

    French Super Intelligence company @MistralAI just released the open weights for Le Chonk, also known as Mistral Large 4…(opens the original post in a new tab)
19. 21:43
    X@hancock_kent· Discussion

    Mistral just released the weights for Mistral Large 4, a 1-trillion-parameter model. Free to use, fine-tune and sell (m…(opens the original post in a new tab)
20. 21:41
    X@Terminator_AI_1· Discussion

    🔴 Mistral open-sources Large 4 with 1.05 trillion parameters French company Mistral AI has publicly released the weigh…(opens the original post in a new tab)
21. 21:30
    X@3li3· Discussion

    رابط الأوزان الرسمية على Hugging Face وتفاصيل الإطلاق من ميسترال: https://mistral.ai/news/mistral-large-4-oss/(opens the original post in a new tab)
22. 21:23
    X@mahadaajani· Discussion

    ☕ 𝗦𝗶𝗽 & 𝗦𝗰𝗿𝗼𝗹𝗹: 🤖 𝗔𝗜 𝗕𝗿𝗶𝗲𝗳𝗶𝗻𝗴 — 𝗙𝗿𝗶𝗱𝗮𝘆, 𝗢𝗰𝘁 𝟯𝟬 A trillion-parameter model goes open, OpenAI shows how it watches its o…(opens the original post in a new tab)
23. 21:20
    X@ErmiyaBM· Discussion

    Mistral open-sourced Large 4 (“Le Chonk”) weights. 1T params, ~50B active. Now tops open models on Terminal-Bench + Co…(opens the original post in a new tab)
24. 21:18
    X@2logics· Discussion

    Mistral AI (@MistralAI) said on Thursday that it has released the open weights of Mistral Large 4, nicknamed "Le Chonk,…(opens the original post in a new tab)
25. 21:17
    X@Mr_Zain72· Discussion

    Mistral just open-sourced Large 4 ("Le Chonk"), a 1T-parameter open-weight MoE model — natively multimodal, reasoning o…(opens the original post in a new tab)
26. 21:16
    X@AINativeF· Discussion

    Mistral AI releases new open-weight model Le Chonk for agentic coding and cybersecurity Mistral AI released the weight…(opens the original post in a new tab)
27. 21:16
    X@nxtday· Discussion

    Mistral Large 4 (Le Chonk) is a 1T-parameter open-weight multimodal model that scores 19.7% on Cognition FrontierCode 2…(opens the original post in a new tab)
28. 21:09
    X@markfenner· Discussion

    Apparently, Le Chonk has been putting in work Mistral reports coding agent performance effectively tied with GLM-5.3 w…(opens the original post in a new tab)
29. 20:33
    X@aiwithbasanth· Discussion

    Mistral just released the open weights of Large 4 ("Le Chonk"): a 1-trillion-parameter multimodal model it calls the be…(opens the original post in a new tab)
30. 20:33
    X@SohaibAnwa11· Discussion

    Sources: https://mistral.ai/news/mistral-large-4-oss/ Independent reporting: https://techcrunch.com/2026/10/29/mistral-…(opens the original post in a new tab)
31. 20:32
    X@SohaibAnwa11· Discussion

    Open weights don't scope an agent's access. Mistral released improved Large 4 weights for coding and cyber defense. I…(opens the original post in a new tab)
32. 20:30
    X@oscarsuiza· Discussion

    Mistral just made its 1 trillion parameter flagship free to download. The open weights of Mistral Large 4, nicknamed L…(opens the original post in a new tab)
33. 20:28
    X@Wu005887· Discussion

    Two signals in one day 👀 🇫🇷 Mistral open-sourced a 1T-param coding model (Le Chonk) 🦑 Google is winding down Jules Mo…(opens the original post in a new tab)
34. 20:26
    X@kilwhissgroup· Discussion

    Mistral has released the weights for Mistral Large 4, a trillion-parameter model for defensive security work. Companies…(opens the original post in a new tab)
35. 20:26
    X@Igoryansky· Discussion

    Mistral released open weights for Mistral Large 4 ("Le Chonk"): ~1T total parameters, 49B active. Each token is routed…(opens the original post in a new tab)
36. 20:25
    X@huang18825· Discussion

    Mistral open-sources the Mistral Large 4 weights, with about 1 trillion total parameters and about 52 billion activated, natively multimodal, focused on code and cybersecurity defense, with API pricing at $0.7 per million input tokens and $2.1 per million output tokens. Europeans have joined the open-source race too?…(opens the original post in a new tab)
37. 20:24
    X@ngocnhat1989· Discussion

    Mistral put the weights of Mistral Large 4, a trillion-parameter model, on Hugging Face and is pitching it for cyber de…(opens the original post in a new tab)
38. 20:24
    X@Shadowfetch· Discussion

    Mistral released the open weights for Mistral Large 4 (Le Chonk), a 1T-parameter multimodal model with a 1M-token conte…(opens the original post in a new tab)
39. 20:21
    X@PXology· Discussion

    Mistral, Large 4 (Le Chonk) modelinin ağırlıklarını yayımladı! 1 trilyon parametreli model artık kurumların kendi alty…(opens the original post in a new tab)
40. 20:19
    X@Rushit_1414· Discussion

    Mistral has released Large 4, a model it calls Le Chonk. Developers can now download its model files; Mistral describes…(opens the original post in a new tab)
41. 20:13
    X@dan3458· Discussion

    Yesterday Mistral open-sourced the weights of the trillion-parameter Large 4 (Le Chonk). Focused on coding agents and cybersecurity, an open model trained in Europe, said to be strong at coding and token-efficient. Who's downloaded it? https://mistral.ai/news/mistral-lar…(opens the original post in a new tab)
42. 20:13
    X@wei_wang· Discussion

    Mistral Large 4 (Le Chonk) has released its weights, with 1T total parameters and about 52B activated per token. For an open model of this scale, how much memory is actually needed at home and how fast it can run — I went through the config and several weight files on Hugging Face…(opens the original post in a new tab)
43. 20:09
    X@DanielZambrini· Discussion

    Good morning! Here is the 30/10/2026 Daily AI digest: Daily AI Digest 30 Oct 2026 X 1/ @MistralAI — The weights for M…(opens the original post in a new tab)
44. 20:06
    Web36Kr· Media coverage

    France's Mistral releases a new-generation AI model targeting the enterprise cybersecurity market(opens the original post in a new tab)
45. 19:56
    X@zuosiojn· Discussion

    Le Chonk is open-sourced: Mistral Large 4, about 1T parameters, weights downloadable and self-hostable. Focused on agentic coding + cybersecurity; first in open-source FrontierCode, saving about 47% tokens at the same coding index tier. Europe…(opens the original post in a new tab)
46. 19:38
    X@XucroYuri· Discussion

    Mistral Large 4: Yesterday Mistral put the weights of Large 4 (nicknamed Le Chonk) on Hugging Face, about 1T total parameters, about 49B activated per token, accepts both images and text, 1M context. You can download it yourself…(opens the original post in a new tab)
47. 19:37
    X@schuer_x· Discussion

    Mistral opened the weights of Mistral Large 4 ("Le Chonk") yesterday: about 1T parameters, 49B active per token. Two k…(opens the original post in a new tab)
48. 19:31
    X@divulgadordeias· Discussion

    Una IA de 1 billón de parámetros… y ahora cualquiera puede descargarla. Mistral liberó Mistral Large 4: según la empre…(opens the original post in a new tab)
49. 19:29
    X@chrimtech· Discussion

    Mistral just let anyone download its biggest AI model: 1 trillion parameters. Mistral says Le Chonk is the best open-w…(opens the original post in a new tab)
50. Time not given
    zhiding.cn

    Mistral open-sources trillion-parameter AI model weights, focusing on the cybersecurity domain

    Mistral released the weights of Mistral Large 4, a trillion-parameter open-source AI model; enterprises and government agencies can download and deploy it themselves, with a focus on cybersecurity. Mistral claims performance comparable to GLM-5.3, 47% fewer Tokens per task, and that it surpasses all other open-source models on the Artificial Analysis agentic coding index. The weights are available on Hugging Face and via API.

51. 01:00
    TechCrunch AI

    Mistral releases the weights of its trillion-parameter model Mistral Large 4 (ML4), targeting cybersecurity

    Mistral releases the weights of its trillion-parameter model Mistral Large 4, which governments and enterprises can deploy themselves for cyber defense. It can locally detect, analyze and respond to threats to code and IT infrastructure. Mistral says performance is comparable to GLM-5.3, using 47% fewer tokens per task, and that it leads all open-source models on the agentic coding tasks of the Coding Agent Index.

1. 23:25
   The Decoder

   Mistral releases Mistral Large 4 ("Le Chonk") weights: leading open-source model in coding, still behind Chinese models overall

   Mistral released the weights of its largest model, Mistral Large 4 ("Le Chonk"), with about 1 trillion parameters and 52B activated per step. Terminal-Bench 4.0 rose from 28% to 41%, slightly above GLM-5.3's 40%, and FrontierCode 2.0 ranked first among open models at 19.7%, but Claude Opus 5.5 scored 46%. It scored only 26 points on the Artificial Analysis Intelligence Index v5, behind DeepSeek V4.1 Flash and GLM-5.3 at 32 points; API pricing is $0.70 per million input tokens and $2.10 per million output tokens.

2. 22:00
   WIRED Security and AI

   Mistral says its AI model Le Chonk has overtaken China in the open-weight race

   Mistral says its trillion-parameter model Le Chonk has become the world's best open-weight model in automated programming and cybersecurity, matching or surpassing models from Chinese labs such as DeepSeek and Z.ai. The model was trained on cybersecurity and programming problems of increasing difficulty, and the final version is now freely available. Security company Sophos cautions that open-weight models can be misused, and any advanced model could go out of control without careful oversight.

3. 20:00
   Mistral AI

   Mistral Large 4 (Le Chonk) open weights: aimed at coding and defensive security

   Mistral AI released the weights of Mistral Large 4 (Le Chonk), a 1 trillion parameter native multimodal model focused on open-source coding and defensive security. It scores 51% on the Artificial Analysis Cyber Index, with an average refusal rate of 96.6% on malicious cyber requests in JailbreakBench, StrongREJECT and AgentHarm.

Current attention 38·Peak in the comparable period 43 (Nov 10 02:00)·Change over the comparable past 24 hours -3%