{"id":110002,"date":"2026-08-08T20:34:07","date_gmt":"2026-08-08T20:34:07","guid":{"rendered":"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/"},"modified":"2026-08-08T20:34:07","modified_gmt":"2026-08-08T20:34:07","slug":"mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size","status":"publish","type":"post","link":"https:\/\/youzum.net\/ja\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/","title":{"rendered":"Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size"},"content":{"rendered":"<p class=\"wp-block-paragraph\">Mistral AI has released <a href=\"https:\/\/mistral.ai\/news\/shieldstral\/\">Shieldstral 1.0 3B<\/a>, an open-weights, policy-adaptive multimodal safety classifier that treats content moderation as a single yes\/no question rather than a fixed taxonomy of harm categories. Most guardrail models bake their category list into the weights, so re-targeting one to a new deployment context means retraining \u2014 and the same content can be acceptable on a cybersecurity research tool while being harmful on a mental-health platform. Shieldstral inverts that: operators write the policy as a plain-language question at inference time, and the model returns a calibrated safety score from a single forward pass. Built on <a href=\"https:\/\/huggingface.co\/mistralai\/Ministral-3-3B-Base-2512\">Ministral-3-3B-Base-2512<\/a> with a native <a href=\"https:\/\/arxiv.org\/abs\/2410.07073\">Pixtral<\/a> vision encoder and released under Apache 2.0, it reports 84.9% average F1 on text safety \u2014 matching GPT-OSS-Safeguard-20B \u2014 and 83.8% on multimodal safety, ahead of every baseline Mistral evaluated.<\/p>\n<h2 class=\"wp-block-heading\"><strong>Is it deployable?<\/strong><\/h2>\n<p class=\"wp-block-paragraph\">Yes, and locally. <a href=\"https:\/\/huggingface.co\/mistralai\/Shieldstral-1.0-3B\">Shieldstral-1.0-3B<\/a> fits in 16GB of VRAM in BF16, runs on a single GPU, and is licensed Apache 2.0 for commercial and non-commercial use. Serving paths are already in place: vLLM (\u22650.26.0, recommended), llama.cpp via GGUF conversion with Q8_0\/Q5_K_M\/Q4_K_M quantization, SGLang, and Transformers \u2014 with fine-tuning supported through <a href=\"https:\/\/github.com\/axolotl-ai-cloud\/axolotl\/tree\/main\/examples\/shieldstral\">Axolotl<\/a>. The classifier emits one token, so latency and cost sit far below reasoning-based guards like GPT-OSS-Safeguard-20B.<\/p>\n<ul class=\"wp-block-list\">\n<li><strong>Which level of company:<\/strong> the 16GB footprint puts it within reach of seed-stage AI product teams that cannot justify a moderation vendor contract, while the open license and self-hosting story suit mid-market and enterprise teams that need guardrails inside a VPC or on-prem for data-residency and audit reasons. Multi-tenant SaaS vendors get a specific win \u2014 one checkpoint can enforce a different policy per customer.<\/li>\n<li><strong>Industries:<\/strong> consumer social and UGC platforms, ed-tech and child-safety surfaces, healthcare and mental-health apps, fintech and insurance support automation, gaming and voice chat, marketplaces and ad\/creative review, and public sector deployments with sovereignty requirements.<\/li>\n<li><strong>Applications:<\/strong> user-prompt moderation, model-response moderation, refusal classification, image-plus-caption review for ads and memes, training-data and RAG-corpus curation, output gating in agentic pipelines, and per-tenant policy enforcement. Because the output is a continuous score rather than a label, teams can tune the threshold per surface or route borderline scores to human review instead of hard-blocking.<\/li>\n<\/ul>\n<h2 class=\"wp-block-heading\"><strong>Moderation as a binary question<\/strong><\/h2>\n<p class=\"wp-block-paragraph\">Shieldstral reduces moderation to one yes\/no question. A fixed system message establishes the task; the user message carries three fields: <code>&lt;Instruct&gt;<\/code> (evaluation context and strictness), <code>&lt;Query&gt;<\/code> (the policy, phrased as a single yes\/no question), and <code>&lt;Document&gt;<\/code> (a prompt, a response, a prompt\u2013response pair, or an image with optional text).<\/p>\n<p class=\"wp-block-paragraph\">At inference the model unembeds only toward the <code>yes<\/code> and <code>no<\/code> token IDs and softmax-normalizes them into a continuous score, thresholded at \u03c4=0.5. That collapses prompt classification, response moderation, refusal detection, and toxicity detection into one problem \u2014 and it means the policy lives entirely in the prompt. Mistral\u2019s guidance is one policy per call; for a broad safe\/unsafe verdict, list the categories in <code>&lt;Instruct&gt;<\/code> and ask a single wide <code>&lt;Query&gt;<\/code>.<\/p>\n<h2 class=\"wp-block-heading\"><strong>The data recipe<\/strong><\/h2>\n<p class=\"wp-block-paragraph\">The claimed advantage comes from data, not scale: roughly <strong>54.1M samples<\/strong> \u2014 45.2M open-source text, 4.4M synthetic contrastive text, 4.5M multimodal. A template-based unification layer converts every dataset into the same instruction\u2013query\u2013document format via per-dataset processors, with randomized phrasings and calibrated strictness (strict for adversarial jailbreaks, lenient for response-quality data).<\/p>\n<p class=\"wp-block-paragraph\">The more interesting piece is contrastive generation. An LLM rewrites safe text into an unsafe variant that violates a target category but deliberately <em>not<\/em> its sibling, producing a positive and a hard negative over identical content in one call. That teaches the model <em>which<\/em> policy is violated rather than a coarse safe\/unsafe split. Image data \u2014 which cannot be synthesized the way text can \u2014 is supplemented with general-purpose image datasets as negatives, query mutation across a 14-subcategory visual taxonomy, and vision\u2013language reranker filtering.<\/p>\n<p class=\"wp-block-paragraph\">Training is LoRA fine-tuning followed by a three-way <a href=\"https:\/\/dl.acm.org\/doi\/10.1145\/325334.325242\">SLERP<\/a> merge: 0.6 public+generated, 0.3 public-only, 0.1 Ministral-3B-Instruct.<\/p>\n<h2 class=\"wp-block-heading\"><strong>Results<\/strong><\/h2>\n<p class=\"wp-block-paragraph\">On text safety, Shieldstral reports <strong>84.9% average F1<\/strong>, tying GPT-OSS-Safeguard-20B (84.9%) as the smallest model in the comparison, with wins on ToxicChat (84.1), HarmBench (99.4), and Aegis v2 response (87.2). On multimodal safety it reports <strong>83.8% overall<\/strong> versus 77.6% for OmniGuard-7B, leading VLGuard (97.7) and UnsafeBench (81.8); LlavaGuard-7B still leads its namesake benchmark at 81.4.<\/p>\n<p class=\"wp-block-paragraph\">On the adaptability benchmark \u2014 built on a deliberately divergent taxonomy of 12 super classes, 26 subcategories, and 52 leaf categories with 90 fixed queries, where no leaf maps one-to-one to training \u2014 Shieldstral scores <strong>91.3% F1<\/strong>, behind GPT-OSS-Safeguard-20B (94.1%) and Nemotron-3.5-Safety-4B (91.8%), but without generating a reasoning trace. Refusal detection lands at 91.5% overall against 93.7% for GPT-OSS-Safeguard-20B.<\/p>\n<p class=\"wp-block-paragraph\"><strong>Where it is weaker:<\/strong> multilingual prompt classification lags on Arabic and Indonesian and on RTP-LX prompts (70.3 vs 86.1 for Nemotron-3.5-Safety-4B). Mistral also flags reduced reliability on adversarial or obfuscated inputs and very long documents. Trained context is 32k tokens across 12 languages.<\/p>\n<div>\n<\/div>\n<p class=\"wp-block-paragraph\">\n<h2 class=\"wp-block-heading\"><strong>Key Takeaways<\/strong><\/h2>\n<ul class=\"wp-block-list\">\n<li>3B Apache 2.0 multimodal guardrail; policy is a plain-language question at inference time, no retraining.<\/li>\n<li>84.9% text F1 ties a 20B model; 83.8% multimodal F1 is best-in-class among evaluated baselines.<\/li>\n<li>54.1M samples with sibling-contrastive rewrites is the actual mechanism behind policy generalization.<\/li>\n<li>Single forward pass, single token out, continuous score at \u03c4=0.5 \u2014 cheap enough for real-time gating.<\/li>\n<li>Weak spots: low-resource languages, obfuscated inputs, long documents.<\/li>\n<\/ul>\n<\/p><p class=\"wp-block-paragraph\">\n<hr class=\"wp-block-separator has-alpha-channel-opacity\" \/>\n<\/p><p class=\"wp-block-paragraph\">\n<\/p><p class=\"wp-block-paragraph\"><strong>Check out the <a href=\"https:\/\/arxiv.org\/abs\/2607.25857\" target=\"_blank\" rel=\"noreferrer noopener\">Paper<\/a>, <a href=\"https:\/\/huggingface.co\/mistralai\/Shieldstral-1.0-3B\" target=\"_blank\" rel=\"noreferrer noopener\">Model on Hugging Face<\/a>, and <a href=\"https:\/\/mistral.ai\/news\/shieldstral\/\" target=\"_blank\" rel=\"noreferrer noopener\">Technical Details<\/a><\/strong>.<strong>\u00a0<\/strong>Also,\u00a0feel free to follow us on\u00a0<strong><a href=\"https:\/\/x.com\/intent\/follow?screen_name=marktechpost\" target=\"_blank\" rel=\"noreferrer noopener\"><mark>Twitter<\/mark><\/a><\/strong>\u00a0and don\u2019t forget to join our\u00a0<strong><a href=\"https:\/\/www.reddit.com\/r\/machinelearningnews\/\" target=\"_blank\" rel=\"noreferrer noopener\">150k+ML SubReddit<\/a><\/strong>\u00a0and Subscribe to\u00a0<strong><a href=\"https:\/\/www.aidevsignals.com\/\" target=\"_blank\" rel=\"noreferrer noopener\">our Newsletter<\/a><\/strong>. Wait! are you on telegram?\u00a0<strong><a href=\"https:\/\/t.me\/machinelearningresearchnews\" target=\"_blank\" rel=\"noreferrer noopener\">now you can join us on telegram as well.<\/a><\/strong><\/p>\n<p class=\"wp-block-paragraph\">Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.?\u00a0<strong><a href=\"https:\/\/forms.gle\/wbash1wF6efRj8G58\" target=\"_blank\" rel=\"noreferrer noopener\"><mark>Connect with us<\/mark><\/a><\/strong><\/p>\n<p>The post <a href=\"https:\/\/www.marktechpost.com\/2026\/08\/07\/mistral-ai-releases-shieldstral-1-0-3b\/\">Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size<\/a> appeared first on <a href=\"https:\/\/www.marktechpost.com\/\">MarkTechPost<\/a>.<\/p>","protected":false},"excerpt":{"rendered":"<p>Mistral AI has released Shieldstral 1.0 3B, an open-weights, policy-adaptive multimodal safety classifier that treats content moderation as a single yes\/no question rather than a fixed taxonomy of harm categories. Most guardrail models bake their category list into the weights, so re-targeting one to a new deployment context means retraining \u2014 and the same content can be acceptable on a cybersecurity research tool while being harmful on a mental-health platform. Shieldstral inverts that: operators write the policy as a plain-language question at inference time, and the model returns a calibrated safety score from a single forward pass. Built on Ministral-3-3B-Base-2512 with a native Pixtral vision encoder and released under Apache 2.0, it reports 84.9% average F1 on text safety \u2014 matching GPT-OSS-Safeguard-20B \u2014 and 83.8% on multimodal safety, ahead of every baseline Mistral evaluated. Is it deployable? Yes, and locally. Shieldstral-1.0-3B fits in 16GB of VRAM in BF16, runs on a single GPU, and is licensed Apache 2.0 for commercial and non-commercial use. Serving paths are already in place: vLLM (\u22650.26.0, recommended), llama.cpp via GGUF conversion with Q8_0\/Q5_K_M\/Q4_K_M quantization, SGLang, and Transformers \u2014 with fine-tuning supported through Axolotl. The classifier emits one token, so latency and cost sit far below reasoning-based guards like GPT-OSS-Safeguard-20B. Which level of company: the 16GB footprint puts it within reach of seed-stage AI product teams that cannot justify a moderation vendor contract, while the open license and self-hosting story suit mid-market and enterprise teams that need guardrails inside a VPC or on-prem for data-residency and audit reasons. Multi-tenant SaaS vendors get a specific win \u2014 one checkpoint can enforce a different policy per customer. Industries: consumer social and UGC platforms, ed-tech and child-safety surfaces, healthcare and mental-health apps, fintech and insurance support automation, gaming and voice chat, marketplaces and ad\/creative review, and public sector deployments with sovereignty requirements. Applications: user-prompt moderation, model-response moderation, refusal classification, image-plus-caption review for ads and memes, training-data and RAG-corpus curation, output gating in agentic pipelines, and per-tenant policy enforcement. Because the output is a continuous score rather than a label, teams can tune the threshold per surface or route borderline scores to human review instead of hard-blocking. Moderation as a binary question Shieldstral reduces moderation to one yes\/no question. A fixed system message establishes the task; the user message carries three fields: &lt;Instruct&gt; (evaluation context and strictness), &lt;Query&gt; (the policy, phrased as a single yes\/no question), and &lt;Document&gt; (a prompt, a response, a prompt\u2013response pair, or an image with optional text). At inference the model unembeds only toward the yes and no token IDs and softmax-normalizes them into a continuous score, thresholded at \u03c4=0.5. That collapses prompt classification, response moderation, refusal detection, and toxicity detection into one problem \u2014 and it means the policy lives entirely in the prompt. Mistral\u2019s guidance is one policy per call; for a broad safe\/unsafe verdict, list the categories in &lt;Instruct&gt; and ask a single wide &lt;Query&gt;. The data recipe The claimed advantage comes from data, not scale: roughly 54.1M samples \u2014 45.2M open-source text, 4.4M synthetic contrastive text, 4.5M multimodal. A template-based unification layer converts every dataset into the same instruction\u2013query\u2013document format via per-dataset processors, with randomized phrasings and calibrated strictness (strict for adversarial jailbreaks, lenient for response-quality data). The more interesting piece is contrastive generation. An LLM rewrites safe text into an unsafe variant that violates a target category but deliberately not its sibling, producing a positive and a hard negative over identical content in one call. That teaches the model which policy is violated rather than a coarse safe\/unsafe split. Image data \u2014 which cannot be synthesized the way text can \u2014 is supplemented with general-purpose image datasets as negatives, query mutation across a 14-subcategory visual taxonomy, and vision\u2013language reranker filtering. Training is LoRA fine-tuning followed by a three-way SLERP merge: 0.6 public+generated, 0.3 public-only, 0.1 Ministral-3B-Instruct. Results On text safety, Shieldstral reports 84.9% average F1, tying GPT-OSS-Safeguard-20B (84.9%) as the smallest model in the comparison, with wins on ToxicChat (84.1), HarmBench (99.4), and Aegis v2 response (87.2). On multimodal safety it reports 83.8% overall versus 77.6% for OmniGuard-7B, leading VLGuard (97.7) and UnsafeBench (81.8); LlavaGuard-7B still leads its namesake benchmark at 81.4. On the adaptability benchmark \u2014 built on a deliberately divergent taxonomy of 12 super classes, 26 subcategories, and 52 leaf categories with 90 fixed queries, where no leaf maps one-to-one to training \u2014 Shieldstral scores 91.3% F1, behind GPT-OSS-Safeguard-20B (94.1%) and Nemotron-3.5-Safety-4B (91.8%), but without generating a reasoning trace. Refusal detection lands at 91.5% overall against 93.7% for GPT-OSS-Safeguard-20B. Where it is weaker: multilingual prompt classification lags on Arabic and Indonesian and on RTP-LX prompts (70.3 vs 86.1 for Nemotron-3.5-Safety-4B). Mistral also flags reduced reliability on adversarial or obfuscated inputs and very long documents. Trained context is 32k tokens across 12 languages. Key Takeaways 3B Apache 2.0 multimodal guardrail; policy is a plain-language question at inference time, no retraining. 84.9% text F1 ties a 20B model; 83.8% multimodal F1 is best-in-class among evaluated baselines. 54.1M samples with sibling-contrastive rewrites is the actual mechanism behind policy generalization. Single forward pass, single token out, continuous score at \u03c4=0.5 \u2014 cheap enough for real-time gating. Weak spots: low-resource languages, obfuscated inputs, long documents. Check out the Paper, Model on Hugging Face, and Technical Details.\u00a0Also,\u00a0feel free to follow us on\u00a0Twitter\u00a0and don\u2019t forget to join our\u00a0150k+ML SubReddit\u00a0and Subscribe to\u00a0our Newsletter. Wait! are you on telegram?\u00a0now you can join us on telegram as well. Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.?\u00a0Connect with us The post Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size appeared first on MarkTechPost.<\/p>","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"pmpro_default_level":"","site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"_pvb_checkbox_block_on_post":false,"footnotes":""},"categories":[52,5,7,1],"tags":[],"class_list":["post-110002","post","type-post","status-publish","format-standard","hentry","category-ai-club","category-committee","category-news","category-uncategorized","pmpro-has-access"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v25.3 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size - YouZum<\/title>\n<meta name=\"description\" content=\"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/youzum.net\/ja\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/\" \/>\n<meta property=\"og:locale\" content=\"ja_JP\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size - YouZum\" \/>\n<meta property=\"og:description\" content=\"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19\" \/>\n<meta property=\"og:url\" content=\"https:\/\/youzum.net\/ja\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/\" \/>\n<meta property=\"og:site_name\" content=\"YouZum\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/DroneAssociationTH\/\" \/>\n<meta property=\"article:published_time\" content=\"2026-08-08T20:34:07+00:00\" \/>\n<meta name=\"author\" content=\"admin NU\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u57f7\u7b46\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin NU\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u63a8\u5b9a\u8aad\u307f\u53d6\u308a\u6642\u9593\" \/>\n\t<meta name=\"twitter:data2\" content=\"5\u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/#article\",\"isPartOf\":{\"@id\":\"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/\"},\"author\":{\"name\":\"admin NU\",\"@id\":\"https:\/\/yousum.gpucore.co\/#\/schema\/person\/97fa48242daf3908e4d9a5f26f4a059c\"},\"headline\":\"Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size\",\"datePublished\":\"2026-08-08T20:34:07+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/\"},\"wordCount\":948,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\/\/yousum.gpucore.co\/#organization\"},\"articleSection\":[\"AI\",\"Committee\",\"News\",\"Uncategorized\"],\"inLanguage\":\"ja\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/\",\"url\":\"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/\",\"name\":\"Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size - YouZum\",\"isPartOf\":{\"@id\":\"https:\/\/yousum.gpucore.co\/#website\"},\"datePublished\":\"2026-08-08T20:34:07+00:00\",\"description\":\"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19\",\"breadcrumb\":{\"@id\":\"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/#breadcrumb\"},\"inLanguage\":\"ja\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/youzum.net\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/yousum.gpucore.co\/#website\",\"url\":\"https:\/\/yousum.gpucore.co\/\",\"name\":\"YouSum\",\"description\":\"\",\"publisher\":{\"@id\":\"https:\/\/yousum.gpucore.co\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/yousum.gpucore.co\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"ja\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/yousum.gpucore.co\/#organization\",\"name\":\"Drone Association Thailand\",\"url\":\"https:\/\/yousum.gpucore.co\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"ja\",\"@id\":\"https:\/\/yousum.gpucore.co\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/youzum.net\/wp-content\/uploads\/2024\/11\/tranparent-logo.png\",\"contentUrl\":\"https:\/\/youzum.net\/wp-content\/uploads\/2024\/11\/tranparent-logo.png\",\"width\":300,\"height\":300,\"caption\":\"Drone Association Thailand\"},\"image\":{\"@id\":\"https:\/\/yousum.gpucore.co\/#\/schema\/logo\/image\/\"},\"sameAs\":[\"https:\/\/www.facebook.com\/DroneAssociationTH\/\"]},{\"@type\":\"Person\",\"@id\":\"https:\/\/yousum.gpucore.co\/#\/schema\/person\/97fa48242daf3908e4d9a5f26f4a059c\",\"name\":\"admin NU\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"ja\",\"@id\":\"https:\/\/yousum.gpucore.co\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/youzum.net\/wp-content\/uploads\/avatars\/2\/1746849356-bpfull.png\",\"contentUrl\":\"https:\/\/youzum.net\/wp-content\/uploads\/avatars\/2\/1746849356-bpfull.png\",\"caption\":\"admin NU\"},\"url\":\"https:\/\/youzum.net\/ja\/members\/adminnu\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size - YouZum","description":"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/youzum.net\/ja\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/","og_locale":"ja_JP","og_type":"article","og_title":"Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size - YouZum","og_description":"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19","og_url":"https:\/\/youzum.net\/ja\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/","og_site_name":"YouZum","article_publisher":"https:\/\/www.facebook.com\/DroneAssociationTH\/","article_published_time":"2026-08-08T20:34:07+00:00","author":"admin NU","twitter_card":"summary_large_image","twitter_misc":{"\u57f7\u7b46\u8005":"admin NU","\u63a8\u5b9a\u8aad\u307f\u53d6\u308a\u6642\u9593":"5\u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/#article","isPartOf":{"@id":"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/"},"author":{"name":"admin NU","@id":"https:\/\/yousum.gpucore.co\/#\/schema\/person\/97fa48242daf3908e4d9a5f26f4a059c"},"headline":"Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size","datePublished":"2026-08-08T20:34:07+00:00","mainEntityOfPage":{"@id":"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/"},"wordCount":948,"commentCount":0,"publisher":{"@id":"https:\/\/yousum.gpucore.co\/#organization"},"articleSection":["AI","Committee","News","Uncategorized"],"inLanguage":"ja","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/","url":"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/","name":"Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size - YouZum","isPartOf":{"@id":"https:\/\/yousum.gpucore.co\/#website"},"datePublished":"2026-08-08T20:34:07+00:00","description":"\u0e01\u0e34\u0e08\u0e01\u0e23\u0e23\u0e21\u0e40\u0e01\u0e35\u0e48\u0e22\u0e27\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e23\u0e19","breadcrumb":{"@id":"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/#breadcrumb"},"inLanguage":"ja","potentialAction":[{"@type":"ReadAction","target":["https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/youzum.net\/mistral-ai-releases-shieldstral-1-0-3b-an-open-weights-policy-adaptive-multimodal-safety-classifier-matching-models-7x-its-size\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/youzum.net\/"},{"@type":"ListItem","position":2,"name":"Mistral AI Releases Shieldstral 1.0 3B: An Open-Weights Policy-Adaptive Multimodal Safety Classifier Matching Models 7\u00d7 Its Size"}]},{"@type":"WebSite","@id":"https:\/\/yousum.gpucore.co\/#website","url":"https:\/\/yousum.gpucore.co\/","name":"YouSum","description":"","publisher":{"@id":"https:\/\/yousum.gpucore.co\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/yousum.gpucore.co\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"ja"},{"@type":"Organization","@id":"https:\/\/yousum.gpucore.co\/#organization","name":"Drone Association Thailand","url":"https:\/\/yousum.gpucore.co\/","logo":{"@type":"ImageObject","inLanguage":"ja","@id":"https:\/\/yousum.gpucore.co\/#\/schema\/logo\/image\/","url":"https:\/\/youzum.net\/wp-content\/uploads\/2024\/11\/tranparent-logo.png","contentUrl":"https:\/\/youzum.net\/wp-content\/uploads\/2024\/11\/tranparent-logo.png","width":300,"height":300,"caption":"Drone Association Thailand"},"image":{"@id":"https:\/\/yousum.gpucore.co\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/DroneAssociationTH\/"]},{"@type":"Person","@id":"https:\/\/yousum.gpucore.co\/#\/schema\/person\/97fa48242daf3908e4d9a5f26f4a059c","name":"admin NU","image":{"@type":"ImageObject","inLanguage":"ja","@id":"https:\/\/yousum.gpucore.co\/#\/schema\/person\/image\/","url":"https:\/\/youzum.net\/wp-content\/uploads\/avatars\/2\/1746849356-bpfull.png","contentUrl":"https:\/\/youzum.net\/wp-content\/uploads\/avatars\/2\/1746849356-bpfull.png","caption":"admin NU"},"url":"https:\/\/youzum.net\/ja\/members\/adminnu\/"}]}},"rttpg_featured_image_url":null,"rttpg_author":{"display_name":"admin NU","author_link":"https:\/\/youzum.net\/ja\/members\/adminnu\/"},"rttpg_comment":0,"rttpg_category":"<a href=\"https:\/\/youzum.net\/ja\/category\/ai-club\/\" rel=\"category tag\">AI<\/a> <a href=\"https:\/\/youzum.net\/ja\/category\/committee\/\" rel=\"category tag\">Committee<\/a> <a href=\"https:\/\/youzum.net\/ja\/category\/news\/\" rel=\"category tag\">News<\/a> <a href=\"https:\/\/youzum.net\/ja\/category\/uncategorized\/\" rel=\"category tag\">Uncategorized<\/a>","rttpg_excerpt":"Mistral AI has released Shieldstral 1.0 3B, an open-weights, policy-adaptive multimodal safety classifier that treats content moderation as a single yes\/no question rather than a fixed taxonomy of harm categories. Most guardrail models bake their category list into the weights, so re-targeting one to a new deployment context means retraining \u2014 and the same content&hellip;","_links":{"self":[{"href":"https:\/\/youzum.net\/ja\/wp-json\/wp\/v2\/posts\/110002","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/youzum.net\/ja\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/youzum.net\/ja\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/youzum.net\/ja\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/youzum.net\/ja\/wp-json\/wp\/v2\/comments?post=110002"}],"version-history":[{"count":0,"href":"https:\/\/youzum.net\/ja\/wp-json\/wp\/v2\/posts\/110002\/revisions"}],"wp:attachment":[{"href":"https:\/\/youzum.net\/ja\/wp-json\/wp\/v2\/media?parent=110002"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/youzum.net\/ja\/wp-json\/wp\/v2\/categories?post=110002"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/youzum.net\/ja\/wp-json\/wp\/v2\/tags?post=110002"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}