{"id":14211,"date":"2026-10-29T18:23:19","date_gmt":"2026-10-29T15:23:19","guid":{"rendered":"https:\/\/heshamsaad.com\/?p=14211"},"modified":"2026-10-29T20:18:24","modified_gmt":"2026-10-29T17:18:24","slug":"mistral-large-4","status":"publish","type":"post","link":"https:\/\/heshamsaad.com\/en\/mistral-large-4\/","title":{"rendered":"Mistral Large 4 Is Now Open Weight: License, Pricing, Benchmarks and What It Means for WordPress"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">Mistral Large 4, the model Mistral nicknames Le Chonk, is now open weight. On October 29, 2026, Mistral released the weights of an improved version of the model it first put into public preview earlier this month. It is a natively multimodal Mixture-of-Experts model with about 1 trillion total parameters, and Mistral says it is the best open-weight model for coding and one of the strongest for defensive cybersecurity.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">This post is not a copy of the announcement. We look at what actually changed since the preview, what the license allows and forbids, what it costs on the API, how the vendor numbers compare with rivals, where the model fits best, and what it means for a WordPress site or a business in the MENA region. Unless we say otherwise, every benchmark below is reported by Mistral, not by an independent lab.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Mistral Large 4 at a glance<\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Size:<\/strong> about 1T total parameters (the docs list 1.05T), with 49B active parameters per token, or 52B including embeddings and output layers.<\/li>\n\n<li><strong>Architecture:<\/strong> granular Mixture-of-Experts with 128 routed experts, 4 active per token, plus 1 shared expert.<\/li>\n\n<li><strong>Input and output:<\/strong> text and image input, text output. The docs mention a 1.6B vision encoder.<\/li>\n\n<li><strong>Context window:<\/strong> 1M tokens.<\/li>\n\n<li><strong>Reasoning:<\/strong> three effort levels, low, medium and high, with high as the default.<\/li>\n\n<li><strong>Languages:<\/strong> Mistral says it is fluent in 160+ languages. Arabic is listed on the Hugging Face model card.<\/li>\n\n<li><strong>License:<\/strong> Mistral Ecosystem Open-Weight License, a modified MIT license.<\/li>\n\n<li><strong>API price:<\/strong> $0.70 per million input tokens, $0.07 per million cached input tokens and $2.10 per million output tokens.<\/li>\n<\/ul>\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"768\" src=\"https:\/\/heshamsaad.com\/wp-content\/uploads\/2026\/10\/071-token-routing-en-1024x768.jpg\" alt=\"How a token is routed in Mistral Large 4: token, router, 128 routed experts, 4 active per token, plus 1 shared expert, text output\" class=\"wp-image-14214\" srcset=\"https:\/\/heshamsaad.com\/wp-content\/uploads\/2026\/10\/071-token-routing-en-1024x768.jpg 1024w, https:\/\/heshamsaad.com\/wp-content\/uploads\/2026\/10\/071-token-routing-en-300x225.jpg 300w, https:\/\/heshamsaad.com\/wp-content\/uploads\/2026\/10\/071-token-routing-en-768x576.jpg 768w, https:\/\/heshamsaad.com\/wp-content\/uploads\/2026\/10\/071-token-routing-en.jpg 1200w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n<h2 class=\"wp-block-heading\">What changed between the preview and the open release<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The preview already existed, so the news is not a new model name. The news is two things: the weights are public, and the model got better after the preview.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Mistral says it spent about three more weeks scaling reinforcement learning on the model. According to its own figures, that moved the model up on several tests:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Terminal-Bench 4.0:<\/strong> 41%, up 13 points from the preview, while using 30% fewer output tokens.<\/li>\n\n<li><strong>FinWorkBench (Finch):<\/strong> 73.3%, up from 67.4% in the preview.<\/li>\n\n<li><strong>BrowseComp:<\/strong> 84.5%, starting from an early SFT checkpoint at 75%.<\/li>\n\n<li><strong>B3 Agent Security Benchmark:<\/strong> resisted 97.1% of attacks, against 93.3% for the preview.<\/li>\n\n<li><strong>Refusal of harmful cyber requests:<\/strong> 97%, up from 95% in the preview. Mistral reports an average of 96.6% on malicious cyber requests.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">The other change is the reasoning menu. The model now has low, medium and high modes. Mistral says low replaces the old None mode for simple tasks, medium is for everyday work and high is for the hardest problems.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">The license: open, with one important condition<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">This is the part business owners should read twice. The license text on Hugging Face is a modified MIT license. It lets you use, copy, modify, fine-tune, distribute and sell copies of the model and build derivative works, for commercial and non-commercial use.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The condition is about Model as a Service. If you or an affiliate operate a Model as a Service business, you need a separate commercial license from Mistral before commercial use. The license defines that as giving a third party access to the model for inference or post-training, for example through an API or hosted endpoint, in a way that gives that party meaningful control over inputs, parameters or training data.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The license also says what does not count:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>End-user products where the model is embedded inside specific features.<\/li>\n\n<li>Simply relaying requests to a model hosted by someone else.<\/li>\n\n<li>Internal use that does not make the model available to third parties.<\/li>\n\n<li>Use through Mistral&#8217;s official products.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">For most agencies and companies, a chatbot on a site or an internal assistant falls into the allowed cases. Reselling raw API access to the model does not. If your plan sits near that line, ask Mistral or a lawyer before you build.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Pricing and access<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Mistral&#8217;s announcement and its model docs give the same API price: $0.70 per million input tokens, $0.07 per million cached input tokens and $2.10 per million output tokens. The API model name is <code>mistral-large-4<\/code>.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The docs list these API features: structured outputs, function calling, document Q&amp;A, prefix, chat completions, batching, agents and conversations, and built-in tools.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Where you can use it, according to Mistral:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Mistral Studio and the Mistral API.<\/li>\n\n<li>OpenRouter.<\/li>\n\n<li>Hugging Face, as open weights, plus a quantized NVFP4 checkpoint for Blackwell GPUs.<\/li>\n\n<li>Microsoft Foundry, with private deployment options for Azure customers.<\/li>\n\n<li>NVIDIA-hosted endpoints and an optimized NVIDIA NIM container.<\/li>\n\n<li>Vibe, Mistral&#8217;s coding agent.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">For self-hosting, the model card recommends vLLM and also lists SGLang, transformers and llama.cpp. Its example vLLM command uses a tensor parallel size of 8, which tells you this is a multi-GPU server model. Fine-tuning guides exist for Axolotl, Unsloth and TRL.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">How it compares with rivals<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">All numbers in this section come from Mistral&#8217;s charts. They are vendor claims, and the rivals were chosen by Mistral. Treat them as a starting point for your own tests.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">On the app side, Google just added <a href=\"https:\/\/heshamsaad.com\/en\/gemini-file-editing\/\">file editing in Gemini<\/a> for docs, PDFs, slides and sheets.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Coding<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Cognition FrontierCode 2.0:<\/strong> Mistral Large 4 19.7%, DeepSeek V4.1 Flash 19.3%, GLM-5.3 17.8%, Kimi K3 17.3%.<\/li>\n\n<li><strong>Artificial Analysis Coding Agent Index:<\/strong> 48%, which Mistral calls effectively tied with GLM-5.3, while using 47% fewer tokens per task on average.<\/li>\n\n<li><strong>Terminal-Bench 4.0:<\/strong> 41%, against 40% for GLM-5.3 in Mistral&#8217;s chart.<\/li>\n\n<li><strong>Surge Code blind human preference:<\/strong> preferred over GLM-5.3, Kimi K3 and DeepSeek V4.1 Flash, and behind Claude Opus 5.5, which won 78% of comparisons against it.<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">Cybersecurity<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">On the Artificial Analysis Cyber Index, Mistral reports 51%. Its own chart shows two open models scoring higher: MiMo-V2.6-Pro at 56% and DeepSeek V4.1 Flash 1017 at 57%. So the honest reading is strong, not first. Where Mistral does claim a lead is safety: the highest refusal rate on malicious cyber requests among the open models it measured.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Work and general intelligence<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Finch:<\/strong> 73.3%, second among open models to Kimi K3 at 77.3%.<\/li>\n\n<li><strong>BrowseComp:<\/strong> 84.5%, above GLM-5.3 at 82.5% and DeepSeek V4.1 Flash 1017 at 83.4%, below Kimi K3 at 91.2%.<\/li>\n\n<li><strong>Artificial Analysis Intelligence Index v5:<\/strong> 26, which the model card calls the highest among western open-weight models. The card also says leading Chinese open-weight models score higher on this index.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">If you followed our coverage of <a href=\"\/en\/kimi-k3-1-pro\/\">Kimi K3.1 Pro<\/a>, Kimi K3 shows up here as the main open rival on general and finance tasks. For closed security models, see our post on <a href=\"\/en\/gemini-4-carbon\/\">Gemini 4 Carbon<\/a>.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Pros and cons<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">Pros<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Open weights you can run on your own hardware or private cloud.<\/li>\n\n<li>A permissive license for normal commercial use.<\/li>\n\n<li>Strong coding results with lower token use, which cuts cost per task.<\/li>\n\n<li>A 1M token context window and image input.<\/li>\n\n<li>Three reasoning levels, so you pay for deep thinking only when you need it.<\/li>\n\n<li>Many access routes, including Microsoft Foundry and NVIDIA.<\/li>\n\n<li>Arabic is among the listed languages.<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">Cons<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Self-hosting a 1T parameter model needs serious multi-GPU servers. For most small businesses the API is the realistic option.<\/li>\n\n<li>Model as a Service businesses need a separate license.<\/li>\n\n<li>Not the top open model on every test: Kimi K3 leads on Finch and BrowseComp, and two models score higher on the Cyber Index.<\/li>\n\n<li>Almost all benchmark numbers are vendor-reported.<\/li>\n\n<li>Text output only, no image or audio generation.<\/li>\n\n<li>Mistral itself says it saw potential uplift for real attacks and is still updating its guardrails.<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">Best use cases<\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Coding agents:<\/strong> fixing bugs, writing features and reviewing pull requests, where token efficiency matters.<\/li>\n\n<li><strong>Security defense:<\/strong> code audits, vulnerability triage and patching, as in Mistral&#8217;s own triage demo.<\/li>\n\n<li><strong>Regulated data:<\/strong> banks, government and health teams that cannot send data to a foreign API can host the model themselves.<\/li>\n\n<li><strong>Long documents:<\/strong> contracts, reports and code bases inside the 1M token window.<\/li>\n\n<li><strong>Finance spreadsheets:<\/strong> Mistral says it improved creating and editing spreadsheets for finance and accounting.<\/li>\n\n<li><strong>Documents and images:<\/strong> reading charts, scanned pages and technical drawings.<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">What Mistral Large 4 means for WordPress sites<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">You will not run Le Chonk on your hosting plan. A WordPress site would call it through the Mistral API or OpenRouter, the same way you call other models. What changes is the choice you have.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Plugin and theme development:<\/strong> a strong coding agent at $0.70 and $2.10 per million tokens is a cheaper option for reviewing PHP, writing blocks and finding bugs.<\/li>\n\n<li><strong>Security checks:<\/strong> the model&#8217;s focus on defensive security fits code audits of custom plugins before deployment. It does not replace updates, backups and a firewall.<\/li>\n\n<li><strong>Content and support:<\/strong> chat assistants, product descriptions and support replies in many languages, with cached input lowering the cost of repeated system prompts.<\/li>\n\n<li><strong>Structured output:<\/strong> JSON schema support makes it easier to feed results into custom fields, WooCommerce data or forms.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Our advice is to test it on your real tasks before switching. Keep your API key on the server, never in front-end JavaScript, and set a monthly spending limit. If you need help wiring a model into your site, our <a href=\"\/en\/wordpress-development\/\">WordPress development<\/a> team can build and test it with you.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">What it means for businesses in MENA<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Data residency is a real issue in the region. Banks, government bodies and health providers in Saudi Arabia, the UAE and Egypt often face rules about where data is stored and processed. An open-weight model that can run in a local data center or private cloud gives these teams an option that a closed API does not.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Sovereign hosting:<\/strong> large organizations can run the model on their own infrastructure and keep data inside the country.<\/li>\n\n<li><strong>Arabic:<\/strong> Arabic is on the language list, but Mistral publishes no Arabic benchmark for this model. Test it on your dialect and your domain before you rely on it.<\/li>\n\n<li><strong>Cost for small teams:<\/strong> for an agency or startup, the API price makes it a reasonable option to compare with other models.<\/li>\n\n<li><strong>Security teams:<\/strong> local security firms can fine-tune the model on their own data, which the license allows.<\/li>\n\n<li><strong>License check:<\/strong> local cloud providers that want to resell access to the model need to talk to Mistral first.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Not sure if it fits your project? Book a <a href=\"\/en\/free-consultation\/\">free consultation<\/a> and we will look at your use case.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Our take<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Mistral Large 4 is a serious open-weight option, especially for coding and security work, at a fair API price. The license is open enough for most businesses. The benchmarks are promising but come from Mistral, and the model is not first everywhere. For WordPress owners, the practical move is simple: try it through the API on a real task, compare cost and quality with your current model, and decide based on your own results.<\/p>\n\n\n<h2 class=\"wp-block-heading\">Frequently asked questions<\/h2>\n\n\n\n<div class=\"wp-block-wpseopress-faq-block-v2 is-layout-flow wp-block-wpseopress-faq-block-v2-is-layout-flow\">\n<details class=\"wp-block-details is-layout-flow wp-block-details-is-layout-flow\" id=\"faq-what-is\"><summary>What is Mistral Large 4?<\/summary>\n<p class=\"wp-block-paragraph\">Mistral Large 4, nicknamed Le Chonk, is Mistral&#8217;s flagship model. It is a natively multimodal Mixture-of-Experts model with about 1 trillion total parameters and 49 billion active per token. It takes text and images, outputs text, has a 1M token context window and offers low, medium and high reasoning modes. Mistral released its weights on October 29, 2026.<\/p>\n<\/details>\n\n<details class=\"wp-block-details is-layout-flow wp-block-details-is-layout-flow\" id=\"faq-license\"><summary>Can I use Mistral Large 4 commercially?<\/summary>\n<p class=\"wp-block-paragraph\">Yes. The Mistral Ecosystem Open-Weight License is a modified MIT license that allows commercial and non-commercial use, fine-tuning and derivative works. The one condition is that businesses operating a Model as a Service offering, meaning they give third parties meaningful control over the model through an API or hosted endpoint, need a separate commercial license from Mistral first.<\/p>\n<\/details>\n\n<details class=\"wp-block-details is-layout-flow wp-block-details-is-layout-flow\" id=\"faq-price\"><summary>How much does Mistral Large 4 cost on the API?<\/summary>\n<p class=\"wp-block-paragraph\">Mistral lists the API price at $0.70 per million input tokens, $0.07 per million cached input tokens and $2.10 per million output tokens. The model name on the API is mistral-large-4. Running the open weights yourself has no license fee for normal use, but you pay for the multi-GPU hardware needed to serve a model of this size.<\/p>\n<\/details>\n\n<details class=\"wp-block-details is-layout-flow wp-block-details-is-layout-flow\" id=\"faq-best\"><summary>Is Mistral Large 4 the best open-weight model?<\/summary>\n<p class=\"wp-block-paragraph\">Not on every test. Mistral reports that it leads open-weight models on FrontierCode 2.0 and Terminal-Bench 4.0 and ties GLM-5.3 on the Coding Agent Index with fewer tokens. In the same charts, Kimi K3 scores higher on Finch and BrowseComp, and two models score higher on the Cyber Index. These are vendor-reported figures.<\/p>\n<\/details>\n\n<details class=\"wp-block-details is-layout-flow wp-block-details-is-layout-flow\" id=\"faq-wordpress\"><summary>How can a WordPress site use Mistral Large 4?<\/summary>\n<p class=\"wp-block-paragraph\">Most sites should call it through the Mistral API or OpenRouter from server-side code or a plugin, not host it. Useful jobs include code review for custom plugins, support chat, product descriptions and structured data through JSON schema output. Keep the API key on the server, set a spending limit and test it on your own tasks first.<\/p>\n<\/details>\n\n<details class=\"wp-block-details is-layout-flow wp-block-details-is-layout-flow\" id=\"faq-arabic\"><summary>Does Mistral Large 4 support Arabic?<\/summary>\n<p class=\"wp-block-paragraph\">Arabic is listed among the supported languages on the Hugging Face model card, and Mistral says the model is fluent in more than 160 languages. However, Mistral has not published Arabic-specific benchmarks for this model, so businesses in the region should test it on their own dialect, terms and content before relying on it in production.<\/p>\n<\/details>\n<\/div>\n<script type=\"application\/ld+json\">{\"@context\":\"https:\/\/schema.org\",\"@type\":\"FAQPage\",\"url\":\"https:\/\/heshamsaad.com\/en\/mistral-large-4\/\",\"@id\":\"https:\/\/heshamsaad.com\/en\/mistral-large-4\/\",\"mainEntity\":[{\"@type\":\"Question\",\"url\":\"https:\/\/heshamsaad.com\/en\/mistral-large-4\/#faq-what-is\",\"name\":\"What is Mistral Large 4?\",\"answerCount\":1,\"acceptedAnswer\":{\"@type\":\"Answer\",\"text\":\"\\u003Cp\\u003EMistral Large 4, nicknamed Le Chonk, is Mistral\\u0027s flagship model. It is a natively multimodal Mixture-of-Experts model with about 1 trillion total parameters and 49 billion active per token. It takes text and images, outputs text, has a 1M token context window and offers low, medium and high reasoning modes. Mistral released its weights on October 29, 2026.\\u003C\/p\\u003E\"}},{\"@type\":\"Question\",\"url\":\"https:\/\/heshamsaad.com\/en\/mistral-large-4\/#faq-license\",\"name\":\"Can I use Mistral Large 4 commercially?\",\"answerCount\":1,\"acceptedAnswer\":{\"@type\":\"Answer\",\"text\":\"\\u003Cp\\u003EYes. The Mistral Ecosystem Open-Weight License is a modified MIT license that allows commercial and non-commercial use, fine-tuning and derivative works. The one condition is that businesses operating a Model as a Service offering, meaning they give third parties meaningful control over the model through an API or hosted endpoint, need a separate commercial license from Mistral first.\\u003C\/p\\u003E\"}},{\"@type\":\"Question\",\"url\":\"https:\/\/heshamsaad.com\/en\/mistral-large-4\/#faq-price\",\"name\":\"How much does Mistral Large 4 cost on the API?\",\"answerCount\":1,\"acceptedAnswer\":{\"@type\":\"Answer\",\"text\":\"\\u003Cp\\u003EMistral lists the API price at $0.70 per million input tokens, $0.07 per million cached input tokens and $2.10 per million output tokens. The model name on the API is mistral-large-4. Running the open weights yourself has no license fee for normal use, but you pay for the multi-GPU hardware needed to serve a model of this size.\\u003C\/p\\u003E\"}},{\"@type\":\"Question\",\"url\":\"https:\/\/heshamsaad.com\/en\/mistral-large-4\/#faq-best\",\"name\":\"Is Mistral Large 4 the best open-weight model?\",\"answerCount\":1,\"acceptedAnswer\":{\"@type\":\"Answer\",\"text\":\"\\u003Cp\\u003ENot on every test. Mistral reports that it leads open-weight models on FrontierCode 2.0 and Terminal-Bench 4.0 and ties GLM-5.3 on the Coding Agent Index with fewer tokens. In the same charts, Kimi K3 scores higher on Finch and BrowseComp, and two models score higher on the Cyber Index. These are vendor-reported figures.\\u003C\/p\\u003E\"}},{\"@type\":\"Question\",\"url\":\"https:\/\/heshamsaad.com\/en\/mistral-large-4\/#faq-wordpress\",\"name\":\"How can a WordPress site use Mistral Large 4?\",\"answerCount\":1,\"acceptedAnswer\":{\"@type\":\"Answer\",\"text\":\"\\u003Cp\\u003EMost sites should call it through the Mistral API or OpenRouter from server-side code or a plugin, not host it. Useful jobs include code review for custom plugins, support chat, product descriptions and structured data through JSON schema output. Keep the API key on the server, set a spending limit and test it on your own tasks first.\\u003C\/p\\u003E\"}},{\"@type\":\"Question\",\"url\":\"https:\/\/heshamsaad.com\/en\/mistral-large-4\/#faq-arabic\",\"name\":\"Does Mistral Large 4 support Arabic?\",\"answerCount\":1,\"acceptedAnswer\":{\"@type\":\"Answer\",\"text\":\"\\u003Cp\\u003EArabic is listed among the supported languages on the Hugging Face model card, and Mistral says the model is fluent in more than 160 languages. However, Mistral has not published Arabic-specific benchmarks for this model, so businesses in the region should test it on their own dialect, terms and content before relying on it in production.\\u003C\/p\\u003E\"}}]}<\/script>\n\n\n<h2 class=\"wp-block-heading\">References<\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a href=\"https:\/\/mistral.ai\/news\/mistral-large-4-oss\/\" rel=\"noopener\" target=\"_blank\">Mistral: Mistral Large 4 (Le Chonk) is now open weight<\/a><\/li>\n\n<li><a href=\"https:\/\/docs.mistral.ai\/models\/mistral-large-4\" rel=\"noopener\" target=\"_blank\">Mistral Docs: Mistral Large 4<\/a><\/li>\n\n<li><a href=\"https:\/\/docs.mistral.ai\/getting-started\/models\/\" rel=\"noopener\" target=\"_blank\">Mistral Docs: Models overview<\/a><\/li>\n\n<li><a href=\"https:\/\/mistral.ai\/pricing\" rel=\"noopener\" target=\"_blank\">Mistral: Pricing<\/a><\/li>\n\n<li><a href=\"https:\/\/huggingface.co\/mistralai\/Mistral-Large-4-1T-A52B\" rel=\"noopener\" target=\"_blank\">Hugging Face: Mistral Large 4 model card<\/a><\/li>\n\n<li><a href=\"https:\/\/huggingface.co\/mistralai\/Mistral-Large-4-1T-A52B\/blob\/main\/LICENSE\" rel=\"noopener\" target=\"_blank\">Hugging Face: Mistral Ecosystem Open-Weight License<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>Mistral released the weights of Mistral Large 4 (Le Chonk). What changed since the preview, the license condition, API pricing, vendor benchmarks against rivals, pros and cons, and what it means for WordPress and MENA businesses.<\/p>\n","protected":false},"author":1611,"featured_media":14213,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"iawp_total_views":0,"footnotes":""},"categories":[569],"tags":[604],"class_list":["post-14211","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-news","tag-ai-news"],"acf":[],"_links":{"self":[{"href":"https:\/\/heshamsaad.com\/en\/wp-json\/wp\/v2\/posts\/14211","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/heshamsaad.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/heshamsaad.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/heshamsaad.com\/en\/wp-json\/wp\/v2\/users\/1611"}],"replies":[{"embeddable":true,"href":"https:\/\/heshamsaad.com\/en\/wp-json\/wp\/v2\/comments?post=14211"}],"version-history":[{"count":1,"href":"https:\/\/heshamsaad.com\/en\/wp-json\/wp\/v2\/posts\/14211\/revisions"}],"predecessor-version":[{"id":14230,"href":"https:\/\/heshamsaad.com\/en\/wp-json\/wp\/v2\/posts\/14211\/revisions\/14230"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/heshamsaad.com\/en\/wp-json\/wp\/v2\/media\/14213"}],"wp:attachment":[{"href":"https:\/\/heshamsaad.com\/en\/wp-json\/wp\/v2\/media?parent=14211"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/heshamsaad.com\/en\/wp-json\/wp\/v2\/categories?post=14211"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/heshamsaad.com\/en\/wp-json\/wp\/v2\/tags?post=14211"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}