{"id":"mist","slug":"mist","name":"MiST","orgId":"dream","family":"mist","releaseDate":null,"availability":"announced","openWeights":true,"license":null,"paramsTotal":30000000000,"paramsSource":"slug","contextWindow":null,"knowledgeCutoff":null,"modelCardUrl":"https://dreamgroup.com/blog/building-cybersecurity-language-models-with-mist","hfRepo":null,"modalities":["text"],"aliases":["MiST","mist-8b","MiST-8B","dreamgroup/MiST-8B","mist-30b-a3b","MiST-30B-A3B","dreamgroup/MiST-30B-A3B"],"aliasNote":null,"origin":"tracked","tracked":{"trackedSince":"2026-10-19","announced":"2026-10-19","blurb":"Dream's open cybersecurity language models (8B and 30B-A3B), with a dedicated security mid-training stage before chat tuning.","api":"none","openWeights":"yes","params":{"text":"8B, and 30B total with about 3B active (MoE)","source":"Dream's Hugging Face model cards"},"facts":[{"label":"Models","value":"MiST-8B (built on Qwen3-8B-Base) and MiST-30B-A3B (built on NVIDIA Nemotron 3.5 Lightning 30B-A3B Base)"},{"label":"Training","value":"Cybersecurity mid-training on about 976M synthetic tokens, then SFT; the 8B adds a DPO step"},{"label":"Focus","value":"Threat intelligence, CVE-to-CWE mapping, MITRE ATT&CK and vulnerability analysis"},{"label":"Status","value":"Research models, separate from Dream's production models; not for autonomous agent use"},{"label":"Paper","value":"MiST: Mid-Training LLMs for Cybersecurity (EMNLP 2026)"},{"label":"License","value":"Apache-2.0 (the 30B-A3B's NVIDIA base materials stay under OpenMDW-1.1)"}],"claims":["Lab-reported scores (self-reported): MiST-8B averages 61.7 across 13 tasks from six public cybersecurity benchmarks, vs 48.6 for Qwen3-8B; MiST-30B-A3B averages 60.3, vs 51.5 for its Nemotron base.","Lab claim: MiST-8B beat a 70B cybersecurity model on Dream's 13-task evaluation."],"sources":[{"kind":"lab","title":"Building Cybersecurity Language Models with MiST (Dream blog)","url":"https://dreamgroup.com/blog/building-cybersecurity-language-models-with-mist","date":"2026-10-19"},{"kind":"lab","title":"Dream announcement on X","url":"https://x.com/DreamGroupAI/status/2112145644743966730","date":"2026-10-19"},{"kind":"repo","title":"MiST-8B on Hugging Face","url":"https://huggingface.co/dreamgroup/MiST-8B","date":"2026-10-19"},{"kind":"repo","title":"MiST-30B-A3B on Hugging Face","url":"https://huggingface.co/dreamgroup/MiST-30B-A3B","date":"2026-10-19"},{"kind":"paper","title":"MiST: Mid-Training LLMs for Cybersecurity","url":"https://arxiv.org/abs/2609.18496","date":"2026-10-19"}],"firstBenchmark":null},"organization":{"id":"dream","name":"Dream","slug":"dream","country":null,"url":null,"color":"#11110F","mark":"D"},"price":null,"answer":{"word":"Pending benchmarks.","scope":"Announced Oct 19, 2026; no independent results yet.","lead":"As of Oct 22, 2026, MiST is pending benchmarks on isitsota: announced Oct 19, 2026, with no independent results on the boards we track yet.","confidence":"low"},"status":"PENDING_BENCHMARKS","verdicts":[]}