{"id":47841,"date":"2026-09-05T13:00:00","date_gmt":"2026-09-05T10:00:00","guid":{"rendered":"https:\/\/ip-calculator.ru\/blog\/?p=47841"},"modified":"2026-08-30T03:35:53","modified_gmt":"2026-08-30T00:35:53","slug":"kejs-qwen38-flash-next-moe","status":"publish","type":"post","link":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/","title":{"rendered":"\u041a\u0435\u0439\u0441: Qwen3.8-Flash-Next \u2014 MoE 125B\/6B \u043d\u0430 DGX Spark"},"content":{"rendered":"<div class=\"fpm_start\"><\/div>\n<p>26 \u0430\u0432\u0433\u0443\u0441\u0442\u0430 2026 \u0433\u043e\u0434\u0430 Alibaba \u0432\u044b\u043f\u0443\u0441\u0442\u0438\u043b\u0430 Qwen3.8-Flash-Next \u2014 \u044d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442\u0430\u043b\u044c\u043d\u0443\u044e \u043c\u043e\u0434\u0435\u043b\u044c \u0441 \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u043e\u0439, \u043a\u043e\u0442\u043e\u0440\u0443\u044e \u043a\u043e\u043c\u0430\u043d\u0434\u0430 \u043d\u0430\u0437\u044b\u0432\u0430\u0435\u0442 \u00ab\u043f\u0440\u0435\u0432\u044c\u044e Qwen4\u00bb. 125 \u043c\u043b\u0440\u0434 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432, \u043d\u043e \u043d\u0430 \u043a\u0430\u0436\u0434\u043e\u043c \u0448\u0430\u0433\u0435 \u0430\u043a\u0442\u0438\u0432\u043d\u044b \u0442\u043e\u043b\u044c\u043a\u043e 6 \u043c\u043b\u0440\u0434. Simon Willison (\u0430\u0432\u0442\u043e\u0440 \u0431\u0438\u0431\u043b\u0438\u043e\u0442\u0435\u043a\u0438 Datasette \u0438 \u043e\u0434\u0438\u043d \u0438\u0437 \u0441\u0430\u043c\u044b\u0445 \u0430\u043a\u0442\u0438\u0432\u043d\u044b\u0445 \u043e\u0431\u043e\u0437\u0440\u0435\u0432\u0430\u0442\u0435\u043b\u0435\u0439 LLM) \u0441\u043a\u0430\u0447\u0430\u043b \u043a\u0432\u0430\u043d\u0442\u0438\u0437\u0438\u0440\u043e\u0432\u0430\u043d\u043d\u044b\u0435 \u0432\u0435\u0441\u0430 \u043e\u0442 Unsloth \u0438 \u0437\u0430\u043f\u0443\u0441\u0442\u0438\u043b \u043d\u0430 DGX Spark.<\/p>\n<p>\u0412 \u044d\u0442\u043e\u0439 \u0441\u0442\u0430\u0442\u044c\u0435 \u2014 \u0440\u0430\u0437\u0431\u043e\u0440 \u043c\u043e\u0434\u0435\u043b\u0438 \u043f\u043e \u043f\u0435\u0440\u0432\u043e\u0438\u0441\u0442\u043e\u0447\u043d\u0438\u043a\u0430\u043c: \u0447\u0442\u043e \u0432\u043d\u0443\u0442\u0440\u0438, \u043a\u0430\u043a \u0440\u0430\u0431\u043e\u0442\u0430\u0435\u0442 MoE 125B\/6B, \u043a\u0430\u043a\u0438\u0435 \u0431\u0435\u043d\u0447\u043c\u0430\u0440\u043a\u0438, \u043a\u0430\u043a \u0441\u043a\u0430\u0447\u0430\u0442\u044c \u0438 \u0437\u0430\u043f\u0443\u0441\u0442\u0438\u0442\u044c, \u0438 \u043f\u043e\u0447\u0435\u043c\u0443 \u044d\u0442\u043e \u0432\u0430\u0436\u043d\u043e \u0434\u043b\u044f \u043f\u043e\u043d\u0438\u043c\u0430\u043d\u0438\u044f \u0442\u0440\u0435\u043d\u0434\u0430 \u00ab\u0431\u043e\u043b\u044c\u0448\u0438\u0435 \u043c\u043e\u0434\u0435\u043b\u0438 \u0441 \u043c\u0430\u043b\u0435\u043d\u044c\u043a\u0438\u043c \u043f\u043e\u0442\u0440\u0435\u0431\u043b\u0435\u043d\u0438\u0435\u043c\u00bb.<\/p>\n<figure><img decoding=\"async\" alt=\"\u0421\u043a\u0440\u0438\u043d\u0448\u043e\u0442 \u043f\u043e\u0441\u0442\u0430 Simon Willison \u043e Qwen3.8-Flash-Next\" src=\"https:\/\/ip-calculator.ru\/blog\/wp-content\/uploads\/2026\/09\/simon-willison-qwen38.png\" \/><figcaption>\u041f\u043e\u0441\u0442 Simon Willison (26.08.2026): \u00abAnother open weights model from Qwen\u2026 125B tokens, but only 6B active which means it gets a significant performance boost.\u00bb \u0418\u0441\u0442\u043e\u0447\u043d\u0438\u043a: simonwillison.net.<\/figcaption><\/figure>\n<h2>\u0427\u0442\u043e \u0442\u0430\u043a\u043e\u0435 Qwen3.8-Flash-Next<\/h2>\n<p>Qwen3.8-Flash-Next \u2014 \u043c\u0443\u043b\u044c\u0442\u0438\u043c\u043e\u0434\u0430\u043b\u044c\u043d\u0430\u044f \u044f\u0437\u044b\u043a\u043e\u0432\u0430\u044f \u043c\u043e\u0434\u0435\u043b\u044c (Causal Language Model with Vision Encoder) \u043e\u0442 Alibaba Qwen Team. \u041b\u0438\u0446\u0435\u043d\u0437\u0438\u044f \u2014 qwen-community-1.0 (\u043e\u0442\u043a\u0440\u044b\u0442\u044b\u0435 \u0432\u0435\u0441\u0430). \u0410\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u0430 \u043e\u0431\u043e\u0437\u043d\u0430\u0447\u0435\u043d\u0430 \u043a\u0430\u043a <code>qwen4exp<\/code> \u2014 \u044d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442\u0430\u043b\u044c\u043d\u0430\u044f \u043e\u0441\u043d\u043e\u0432\u0430 \u0434\u043b\u044f \u0431\u0443\u0434\u0443\u0449\u0435\u0433\u043e \u043f\u043e\u043a\u043e\u043b\u0435\u043d\u0438\u044f Qwen4.<\/p>\n<p>\u041a\u043b\u044e\u0447\u0435\u0432\u044b\u0435 \u0445\u0430\u0440\u0430\u043a\u0442\u0435\u0440\u0438\u0441\u0442\u0438\u043a\u0438 \u043f\u043e \u0434\u0430\u043d\u043d\u044b\u043c \u043a\u0430\u0440\u0442\u043e\u0447\u043a\u0438 \u043c\u043e\u0434\u0435\u043b\u0438 \u043d\u0430 Hugging Face (https:\/\/huggingface.co\/Qwen\/Qwen3.8-Flash-Next):<\/p>\n<ul>\n<li><strong>\u041f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u044b:<\/strong> 125 \u043c\u043b\u0440\u0434 total, 6 \u043c\u043b\u0440\u0434 \u0430\u043a\u0442\u0438\u0432\u043d\u044b\u0445 \u043d\u0430 \u043a\u0430\u0436\u0434\u043e\u043c \u0448\u0430\u0433\u0435.<\/li>\n<li><strong>N-gram Embedding:<\/strong> 51 \u043c\u043b\u0440\u0434 \u0434\u043e\u043f\u043e\u043b\u043d\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u0445 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432 (\u0431\u0438\u0433\u0440\u0430\u043c\u043c\u044b\/\u0442\u0440\u0438\u0433\u0440\u0430\u043c\u043c\u044b).<\/li>\n<li><strong>MTP (Multi-Token Prediction):<\/strong> 4 \u043c\u043b\u0440\u0434 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432, 1 \u0441\u043b\u043e\u0439.<\/li>\n<li><strong>\u041a\u043e\u043d\u0442\u0435\u043a\u0441\u0442:<\/strong> 262 144 \u0442\u043e\u043a\u0435\u043d\u0430 \u043d\u0430\u0442\u0438\u0432\u043d\u043e, \u0434\u043e 1 000 000 \u0447\u0435\u0440\u0435\u0437 RoPE-\u044d\u043a\u0441\u0442\u0440\u0430\u043f\u043e\u043b\u044f\u0446\u0438\u044e (YaRN).<\/li>\n<li><strong>\u0412\u0432\u043e\u0434:<\/strong> \u0442\u0435\u043a\u0441\u0442 + \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u044f + \u0432\u0438\u0434\u0435\u043e (\u043c\u0443\u043b\u044c\u0442\u0438\u043c\u043e\u0434\u0430\u043b\u044c\u043d\u0430\u044f).<\/li>\n<li><strong>\u042d\u043a\u0441\u043f\u0435\u0440\u0442\u044b MoE:<\/strong> 512, \u0430\u043a\u0442\u0438\u0432\u0438\u0440\u0443\u044e\u0442\u0441\u044f 10 \u043c\u0430\u0440\u0448\u0440\u0443\u0442\u0438\u0437\u0438\u0440\u043e\u0432\u0430\u043d\u043d\u044b\u0445 + 1 \u043e\u0431\u0449\u0438\u0439.<\/li>\n<\/ul>\n<p>\u041a\u0440\u043e\u043c\u0435 \u043e\u0442\u043a\u0440\u044b\u0442\u044b\u0445 \u0432\u0435\u0441\u043e\u0432, \u0434\u043e\u0441\u0442\u0443\u043f\u043d\u0430 \u043e\u0431\u043b\u0430\u0447\u043d\u0430\u044f \u0432\u0435\u0440\u0441\u0438\u044f Qwen3.8-Flash \u0447\u0435\u0440\u0435\u0437 Qwen Cloud (https:\/\/www.qwencloud.com) \u2014 \u043f\u0440\u043e\u0434\u0430\u043a\u0448\u043d-\u0432\u0430\u0440\u0438\u0430\u043d\u0442 \u0441 1M \u043a\u043e\u043d\u0442\u0435\u043a\u0441\u0442\u0430 \u0438 \u0432\u0441\u0442\u0440\u043e\u0435\u043d\u043d\u044b\u043c\u0438 \u0438\u043d\u0441\u0442\u0440\u0443\u043c\u0435\u043d\u0442\u0430\u043c\u0438. \u041d\u043e \u043d\u0430\u0441 \u0438\u043d\u0442\u0435\u0440\u0435\u0441\u0443\u0435\u0442 \u043b\u043e\u043a\u0430\u043b\u044c\u043d\u044b\u0439 \u0437\u0430\u043f\u0443\u0441\u043a.<\/p>\n<figure><img decoding=\"async\" alt=\"\u041a\u0430\u0440\u0442\u043e\u0447\u043a\u0430 Qwen3.8-Flash-Next \u043d\u0430 Hugging Face\" src=\"https:\/\/ip-calculator.ru\/blog\/wp-content\/uploads\/2026\/09\/hf-qwen38-flash-next-48423.png\" \/><figcaption>\u041a\u0430\u0440\u0442\u043e\u0447\u043a\u0430 \u043c\u043e\u0434\u0435\u043b\u0438 Qwen\/Qwen3.8-Flash-Next \u043d\u0430 Hugging Face: \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u0430 qwen4exp, 188K \u0441\u043a\u0430\u0447\u0438\u0432\u0430\u043d\u0438\u0439 \u0437\u0430 \u043c\u0435\u0441\u044f\u0446. \u0418\u0441\u0442\u043e\u0447\u043d\u0438\u043a: huggingface.co\/Qwen\/Qwen3.8-Flash-Next.<\/figcaption><\/figure>\n<h2>\u0410\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u0430: \u043a\u0430\u043a 125 \u043c\u043b\u0440\u0434 \u0441\u0442\u0430\u043d\u043e\u0432\u044f\u0442\u0441\u044f 6 \u043c\u043b\u0440\u0434<\/h2>\n<p>\u041a\u043b\u0430\u0441\u0441\u0438\u0447\u0435\u0441\u043a\u0438\u0435 \u043c\u043e\u0434\u0435\u043b\u0438 \u0440\u0430\u0431\u043e\u0442\u0430\u044e\u0442 \u0441\u043e \u0432\u0441\u0435\u043c\u0438 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u0430\u043c\u0438 \u043d\u0430 \u043a\u0430\u0436\u0434\u043e\u043c \u0448\u0430\u0433\u0435: 7B \u2014 \u044d\u0442\u043e 7B \u0432\u044b\u0447\u0438\u0441\u043b\u0435\u043d\u0438\u0439, 27B \u2014 27B. MoE (Mixture of Experts) \u0440\u0430\u0437\u0431\u0438\u0432\u0430\u0435\u0442 \u0441\u043b\u043e\u0439 \u043d\u0430 \u043c\u043d\u043e\u0436\u0435\u0441\u0442\u0432\u043e \u00ab\u044d\u043a\u0441\u043f\u0435\u0440\u0442\u043e\u0432\u00bb \u0438 \u0430\u043a\u0442\u0438\u0432\u0438\u0440\u0443\u0435\u0442 \u0442\u043e\u043b\u044c\u043a\u043e \u0447\u0430\u0441\u0442\u044c \u0438\u0437 \u043d\u0438\u0445 \u0447\u0435\u0440\u0435\u0437 \u043c\u0430\u0440\u0448\u0440\u0443\u0442\u0438\u0437\u0430\u0442\u043e\u0440.<\/p>\n<p>\u0412 Qwen3.8-Flash-Next \u044d\u0442\u043e \u0440\u0435\u0430\u043b\u0438\u0437\u043e\u0432\u0430\u043d\u043e \u0447\u0435\u0440\u0435\u0437 48 \u0441\u043b\u043e\u0451\u0432. \u041a\u0430\u0436\u0434\u044b\u0439 \u0441\u043b\u043e\u0439 \u0441\u043e\u0434\u0435\u0440\u0436\u0438\u0442 3 \u0431\u043b\u043e\u043a\u0430 Gated DeltaNet \u2192 MoE \u0438 1 \u0431\u043b\u043e\u043a Qwen Sparse Attention (QSA) \u2192 MoE. \u0412\u0441\u0435\u0433\u043e 512 \u044d\u043a\u0441\u043f\u0435\u0440\u0442\u043e\u0432, \u043d\u043e \u043d\u0430 \u043a\u0430\u0436\u0434\u043e\u043c \u0442\u043e\u043a\u0435\u043d\u0435 \u0440\u0430\u0431\u043e\u0442\u0430\u044e\u0442 \u0432\u0441\u0435\u0433\u043e 11 (10 \u043c\u0430\u0440\u0448\u0440\u0443\u0442\u0438\u0437\u0438\u0440\u043e\u0432\u0430\u043d\u043d\u044b\u0445 + 1 \u043e\u0431\u0449\u0438\u0439).<\/p>\n<figure><img decoding=\"async\" alt=\"\u0421\u0445\u0435\u043c\u0430: MoE 125B\/6B \u2014 \u043a\u0430\u043a \u0440\u0430\u0431\u043e\u0442\u0430\u044e\u0442 \u0430\u043a\u0442\u0438\u0432\u043d\u044b\u0435 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u044b\" src=\"https:\/\/ip-calculator.ru\/blog\/wp-content\/uploads\/2026\/09\/moe-architecture.png\" \/><figcaption>\u0410\u0432\u0442\u043e\u0440\u0441\u043a\u0430\u044f \u0441\u0445\u0435\u043c\u0430: \u0438\u0437 512 \u044d\u043a\u0441\u043f\u0435\u0440\u0442\u043e\u0432 \u043d\u0430 \u043a\u0430\u0436\u0434\u044b\u0439 \u0442\u043e\u043a\u0435\u043d \u0430\u043a\u0442\u0438\u0432\u043d\u044b \u0442\u043e\u043b\u044c\u043a\u043e 11. N-gram Embedding \u0434\u043e\u0431\u0430\u0432\u043b\u044f\u0435\u0442 51 \u043c\u043b\u0440\u0434 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432 \u0431\u0435\u0437 \u0432\u044b\u0447\u0438\u0441\u043b\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u0445 \u0437\u0430\u0442\u0440\u0430\u0442.<\/figcaption><\/figure>\n<p>\u041f\u043e\u043c\u0438\u043c\u043e MoE, \u0432 \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u0435 \u0442\u0440\u0438 \u043a\u043b\u044e\u0447\u0435\u0432\u044b\u0445 \u043d\u043e\u0432\u0448\u0435\u0441\u0442\u0432\u0430:<\/p>\n<p><strong>Hybrid Attention \u0441 QSA (Qwen Sparse Attention).<\/strong> \u0412\u043c\u0435\u0441\u0442\u043e \u043a\u043b\u0430\u0441\u0441\u0438\u0447\u0435\u0441\u043a\u043e\u0433\u043e \u043c\u0435\u0445\u0430\u043d\u0438\u0437\u043c\u0430 \u0432\u043d\u0438\u043c\u0430\u043d\u0438\u044f, \u043a\u043e\u0442\u043e\u0440\u044b\u0439 \u043e\u0431\u0440\u0430\u0431\u0430\u0442\u044b\u0432\u0430\u0435\u0442 \u043a\u0430\u0436\u0434\u044b\u0439 \u0442\u043e\u043a\u0435\u043d, QSA \u0440\u0430\u0431\u043e\u0442\u0430\u0435\u0442 \u043d\u0430 \u0443\u0440\u043e\u0432\u043d\u0435 \u043c\u0438\u043a\u0440\u043e\u0431\u043b\u043e\u043a\u043e\u0432. \u042d\u0442\u043e \u0441\u043d\u0438\u0436\u0430\u0435\u0442 \u043b\u0430\u0442\u0435\u043d\u0442\u043d\u043e\u0441\u0442\u044c \u043f\u0440\u0438 \u0434\u043b\u0438\u043d\u043d\u043e\u043c \u043a\u043e\u043d\u0442\u0435\u043a\u0441\u0442\u0435 \u2014 \u043a\u0440\u0438\u0442\u0438\u0447\u043d\u043e \u0434\u043b\u044f \u0430\u0433\u0435\u043d\u0442\u043d\u044b\u0445 \u0441\u0446\u0435\u043d\u0430\u0440\u0438\u0435\u0432, \u0433\u0434\u0435 \u043c\u043e\u0434\u0435\u043b\u044c \u043e\u0431\u0440\u0430\u0431\u0430\u0442\u044b\u0432\u0430\u0435\u0442 \u0434\u0435\u0441\u044f\u0442\u043a\u0438 \u0442\u044b\u0441\u044f\u0447 \u0442\u043e\u043a\u0435\u043d\u043e\u0432 \u0437\u0430 \u0445\u043e\u0434.<\/p>\n<p><strong>Gated Residual.<\/strong> \u0428\u0438\u0440\u043e\u043a\u0438\u0435 residual streams \u0441 data-dependent read gate \u0438 per-branch scalar write gate. \u041a\u043b\u0430\u0441\u0441\u0438\u0447\u0435\u0441\u043a\u0438\u0435 residual connections \u0441 \u043d\u043e\u0440\u043c\u0430\u043b\u0438\u0437\u0430\u0446\u0438\u0435\u0439 \u0434\u0435\u043b\u0430\u044e\u0442 \u0433\u043b\u0443\u0431\u043e\u043a\u043e\u0435 \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u0435 \u0443\u043f\u0440\u0430\u0432\u043b\u044f\u0435\u043c\u044b\u043c; Gated Residual \u0434\u043e\u0431\u0430\u0432\u043b\u044f\u0435\u0442 \u0431\u043e\u043b\u0435\u0435 \u0442\u043e\u043d\u043a\u0443\u044e \u0440\u0435\u0433\u0443\u043b\u0438\u0440\u043e\u0432\u043a\u0443 \u043f\u043e\u0442\u043e\u043a\u043e\u0432 \u0438\u043d\u0444\u043e\u0440\u043c\u0430\u0446\u0438\u0438 \u043c\u0435\u0436\u0434\u0443 \u0441\u043b\u043e\u044f\u043c\u0438 \u0431\u0435\u0437 \u0443\u0432\u0435\u043b\u0438\u0447\u0435\u043d\u0438\u044f \u043d\u0430\u043a\u043b\u0430\u0434\u043d\u044b\u0445 \u0440\u0430\u0441\u0445\u043e\u0434\u043e\u0432 \u043d\u0430 \u0438\u043d\u0444\u0435\u0440\u0435\u043d\u0441.<\/p>\n<p><strong>N-gram Embedding.<\/strong> 51 \u043c\u043b\u0440\u0434 \u0434\u043e\u043f\u043e\u043b\u043d\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u0445 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432, \u043a\u043e\u0442\u043e\u0440\u044b\u0435 \u043d\u0435 \u0443\u0447\u0430\u0441\u0442\u0432\u0443\u044e\u0442 \u0432 \u0432\u044b\u0447\u0438\u0441\u043b\u0435\u043d\u0438\u044f\u0445 \u2014 \u043e\u043d\u0438 indexing-based. \u0411\u0438\u0433\u0440\u0430\u043c\u043c\u044b \u0438 \u0442\u0440\u0438\u0433\u0440\u0430\u043c\u043c\u044b \u043d\u0430 \u0432\u0442\u043e\u0440\u043e\u043c \u0441\u043b\u043e\u0435 \u043f\u043e\u0437\u0432\u043e\u043b\u044f\u044e\u0442 \u043c\u0430\u0441\u0448\u0442\u0430\u0431\u0438\u0440\u043e\u0432\u0430\u0442\u044c \u00ab\u0437\u043d\u0430\u043d\u0438\u0435\u00bb \u043c\u043e\u0434\u0435\u043b\u0438 \u0447\u0435\u0440\u0435\u0437 \u043f\u0430\u043c\u044f\u0442\u044c, \u0430 \u043d\u0435 \u0447\u0435\u0440\u0435\u0437 \u0432\u044b\u0447\u0438\u0441\u043b\u0435\u043d\u0438\u044f. \u042d\u0442\u043e \u0434\u0435\u0448\u0435\u0432\u043b\u0435 MoE \u0434\u043b\u044f memory-constrained \u0443\u0441\u043a\u043e\u0440\u0438\u0442\u0435\u043b\u0435\u0439.<\/p>\n<h2>\u0411\u0435\u043d\u0447\u043c\u0430\u0440\u043a\u0438: \u0446\u0438\u0444\u0440\u044b \u043f\u0440\u043e\u0442\u0438\u0432 \u0444\u043b\u0430\u0433\u043c\u0430\u043d\u043e\u0432<\/h2>\n<p>\u0420\u0435\u0437\u0443\u043b\u044c\u0442\u0430\u0442\u044b \u0441 \u043a\u0430\u0440\u0442\u043e\u0447\u043a\u0438 \u043c\u043e\u0434\u0435\u043b\u0438 \u043d\u0430 Hugging Face. \u0412\u0441\u0435 \u043c\u043e\u0434\u0435\u043b\u0438 \u043e\u0446\u0435\u043d\u0435\u043d\u044b \u0432 \u0438\u044e\u043b\u0435\u2013\u0430\u0432\u0433\u0443\u0441\u0442\u0435 2026:<\/p>\n<table>\n<thead>\n<tr>\n<th>\u0417\u0430\u0434\u0430\u0447\u0430<\/th>\n<th>Qwen3.8-Flash-Next (125B\/6B)<\/th>\n<th>Qwen 3.8 27B<\/th>\n<th>DeepSeek-V4-Flash<\/th>\n<th>Claude Opus 4.6 Max<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>SWE-bench Pro (\u043a\u043e\u0434\u0438\u043d\u0433)<\/td>\n<td><strong>62.5<\/strong><\/td>\n<td>61.7<\/td>\n<td>56.0<\/td>\n<td>53.4<\/td>\n<\/tr>\n<tr>\n<td>DeepSWE 1.1 (\u0430\u0433\u0435\u043d\u0442\u043d\u044b\u0439 \u043a\u043e\u0434\u0438\u043d\u0433)<\/td>\n<td><strong>58.7<\/strong><\/td>\n<td>42.2<\/td>\n<td>54.4<\/td>\n<td>\u2014<\/td>\n<\/tr>\n<tr>\n<td>SWE-bench Multilingual<\/td>\n<td><strong>81.0<\/strong><\/td>\n<td>73.8<\/td>\n<td>\u2014<\/td>\n<td>77.5<\/td>\n<\/tr>\n<tr>\n<td>LiveCodeBench v6 (\u043a\u043e\u043d\u043a\u0443\u0440\u0435\u043d\u0442\u043d\u044b\u0439 \u043a\u043e\u0434\u0438\u043d\u0433)<\/td>\n<td><strong>91.9<\/strong><\/td>\n<td>90.3<\/td>\n<td>90.6<\/td>\n<td>88.8<\/td>\n<\/tr>\n<tr>\n<td>GPQA Diamond (\u043d\u0430\u0443\u0447\u043d\u044b\u0435 \u0440\u0430\u0441\u0441\u0443\u0436\u0434\u0435\u043d\u0438\u044f)<\/td>\n<td>91.7<\/td>\n<td>89.2<\/td>\n<td>90.8<\/td>\n<td>91.3<\/td>\n<\/tr>\n<tr>\n<td>CoWorkBench (\u043e\u0444\u0438\u0441\u043d\u044b\u0435 \u0437\u0430\u0434\u0430\u0447\u0438)<\/td>\n<td><strong>73.9<\/strong><\/td>\n<td>70.7<\/td>\n<td>45.1<\/td>\n<td>68.2<\/td>\n<\/tr>\n<tr>\n<td>IFBench (\u0441\u043b\u0435\u0434\u043e\u0432\u0430\u043d\u0438\u0435 \u0438\u043d\u0441\u0442\u0440\u0443\u043a\u0446\u0438\u044f\u043c)<\/td>\n<td><strong>81.3<\/strong><\/td>\n<td>79.5<\/td>\n<td>79.2<\/td>\n<td>62.5<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Qwen3.8-Flash-Next \u0441 6 \u043c\u043b\u0440\u0434 \u0430\u043a\u0442\u0438\u0432\u043d\u044b\u0445 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432 \u043e\u0431\u0445\u043e\u0434\u0438\u0442 Claude Opus 4.6 Max \u043d\u0430 SWE-bench Pro (62.5 vs 53.4) \u2014 \u0437\u0430\u0434\u0430\u0447\u0430\u0445 \u043f\u043e \u0438\u0441\u043f\u0440\u0430\u0432\u043b\u0435\u043d\u0438\u044e \u0431\u0430\u0433\u043e\u0432 \u0432 \u0440\u0435\u0430\u043b\u044c\u043d\u044b\u0445 \u0440\u0435\u043f\u043e\u0437\u0438\u0442\u043e\u0440\u0438\u044f\u0445. \u041d\u0430 LiveCodeBench v6 \u2014 91.9 \u043f\u0440\u043e\u0442\u0438\u0432 88.8. \u0420\u0430\u0437\u043d\u0438\u0446\u0430 \u0432 62 \u0440\u0430\u0437\u0430 \u043f\u043e \u0447\u0438\u0441\u043b\u0443 \u0430\u043a\u0442\u0438\u0432\u043d\u044b\u0445 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432 (6 \u043c\u043b\u0440\u0434 vs ~400+ \u043c\u043b\u0440\u0434 \u0443 \u0444\u043b\u0430\u0433\u043c\u0430\u043d\u043e\u0432), \u0430 \u0440\u0435\u0437\u0443\u043b\u044c\u0442\u0430\u0442\u044b \u0441\u043e\u043f\u043e\u0441\u0442\u0430\u0432\u0438\u043c\u044b \u0438\u043b\u0438 \u043b\u0443\u0447\u0448\u0435.<\/p>\n<p>\u041a\u043e\u043d\u0435\u0447\u043d\u043e, \u0431\u0435\u043d\u0447\u043c\u0430\u0440\u043a\u0438 \u2014 \u043d\u0435 \u0432\u0441\u044f \u043a\u0430\u0440\u0442\u0438\u043d\u0430. \u041d\u0430 HLE (Hard Likelihood Evaluation) Claude Opus 4.6 Max \u0432\u0441\u0451 \u0435\u0449\u0451 \u043b\u0438\u0434\u0438\u0440\u0443\u0435\u0442 (40.0 vs 35.9). \u041d\u043e \u0434\u043b\u044f \u043a\u043e\u0434\u0438\u043d\u0433\u0430 \u2014 MoE-\u043f\u043e\u0434\u0445\u043e\u0434 \u043f\u043e\u043a\u0430\u0437\u044b\u0432\u0430\u0435\u0442 Pareto-\u043f\u0440\u0435\u0438\u043c\u0443\u0449\u0435\u0441\u0442\u0432\u043e.<\/p>\n<h2>\u041e\u043f\u044b\u0442 Simon Willison: DGX Spark + Unsloth-\u043a\u0432\u0430\u043d\u0442\u044b<\/h2>\n<p>Simon Willison (https:\/\/simonwillison.net\/2026\/Aug\/26\/qwen38-flash-next\/) \u043f\u0440\u043e\u0442\u0435\u0441\u0442\u0438\u0440\u043e\u0432\u0430\u043b \u043c\u043e\u0434\u0435\u043b\u044c \u043d\u0430 DGX Spark \u2014 \u043a\u043e\u043c\u043f\u0430\u043a\u0442\u043d\u043e\u043c \u0441\u0435\u0440\u0432\u0435\u0440\u0435 \u043e\u0442 NVIDIA \u0441 unified memory (48 \u0413\u0411) \u0441\u0442\u043e\u0438\u043c\u043e\u0441\u0442\u044c\u044e \u043e\u043a\u043e\u043b\u043e $3 000.<\/p>\n<p>\u0414\u043b\u044f \u0437\u0430\u043f\u0443\u0441\u043a\u0430 \u043e\u043d \u0438\u0441\u043f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u043b \u043a\u0432\u0430\u043d\u0442\u0438\u0437\u0438\u0440\u043e\u0432\u0430\u043d\u043d\u044b\u0435 GGUF-\u0432\u0435\u0440\u0441\u0438\u0438 \u043e\u0442 Unsloth (https:\/\/huggingface.co\/unsloth\/Qwen3.8-Flash-Next-GGUF):<\/p>\n<ul>\n<li><strong>UD-IQ1_S<\/strong> \u2014 72.5 \u0413\u0411. \u0421\u0430\u043c\u044b\u0439 \u043a\u043e\u043c\u043f\u0430\u043a\u0442\u043d\u044b\u0439 \u043a\u0432\u0430\u043d\u0442. Simon \u0433\u0435\u043d\u0435\u0440\u0438\u0440\u043e\u0432\u0430\u043b SVG-\u0438\u043b\u043b\u044e\u0441\u0442\u0440\u0430\u0446\u0438\u0438 (\u043f\u0435\u043b\u0438\u043a\u0430\u043d\u044b \u043d\u0430 \u0432\u0435\u043b\u043e\u0441\u0438\u043f\u0435\u0434\u0435), \u043c\u043e\u0434\u0435\u043b\u044c \u0441\u043f\u0440\u0430\u0432\u043b\u044f\u043b\u0430\u0441\u044c, \u043d\u043e \u043f\u0440\u0438 1-\u0431\u0438\u0442\u043d\u043e\u043c \u043a\u0432\u0430\u043d\u0442\u043e\u0432\u0430\u043d\u0438\u0438 \u043a\u0430\u0447\u0435\u0441\u0442\u0432\u043e \u0437\u0430\u043c\u0435\u0442\u043d\u043e \u0434\u0435\u0433\u0440\u0430\u0434\u0438\u0440\u0443\u0435\u0442.<\/li>\n<li><strong>UD-Q2_K_XL<\/strong> \u2014 78.9 \u0413\u0411. Simon \u043d\u0430\u0437\u0432\u0430\u043b \u0435\u0433\u043e \u0444\u0430\u0432\u043e\u0440\u0438\u0442\u043e\u043c. \u041d\u0430 DGX Spark \u0441 \u0432\u044b\u0433\u0440\u0443\u0437\u043a\u043e\u0439 (offloading) \u0447\u0430\u0441\u0442\u0438 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432 \u043d\u0430 NVMe \u2014 \u0440\u0430\u0431\u043e\u0442\u0430\u0435\u0442, \u0433\u0435\u043d\u0435\u0440\u0430\u0446\u0438\u044f \u043f\u0440\u0438\u0435\u043c\u043b\u0435\u043c\u0430\u044f.<\/li>\n<\/ul>\n<blockquote>\n<p>My favorite so far was this xhigh reasoning effort one from UD-Q2_K_XL.<\/p>\n<p>\u2014 Simon Willison (https:\/\/simonwillison.net\/2026\/Aug\/26\/qwen38-flash-next\/), 26.08.2026<\/p>\n<\/blockquote>\n<p>\u0412\u0430\u0436\u043d\u044b\u0439 \u043d\u044e\u0430\u043d\u0441: DGX Spark \u0438\u043c\u0435\u0435\u0442 48 \u0413\u0411 unified memory, \u0430 \u043a\u0432\u0430\u043d\u0442\u044b \u0432\u0435\u0441\u044f\u0442 72\u201379 \u0413\u0411. \u0414\u043b\u044f \u0437\u0430\u043f\u0443\u0441\u043a\u0430 \u0442\u0440\u0435\u0431\u0443\u0435\u0442\u0441\u044f \u0432\u044b\u0433\u0440\u0443\u0437\u043a\u0430 \u0447\u0430\u0441\u0442\u0438 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432 \u043d\u0430 \u0434\u0438\u0441\u043a \u2014 \u044d\u0442\u043e \u0441\u043d\u0438\u0436\u0430\u0435\u0442 \u0441\u043a\u043e\u0440\u043e\u0441\u0442\u044c. \u041f\u043e\u043b\u043d\u043e\u0446\u0435\u043d\u043d\u044b\u0439 \u0438\u043d\u0444\u0435\u0440\u0435\u043d\u0441 \u0432\u043e\u0437\u043c\u043e\u0436\u0435\u043d \u043d\u0430 \u0441\u0435\u0440\u0432\u0435\u0440\u0430\u0445 \u0441 96+ \u0413\u0411 VRAM.<\/p>\n<h2>\u041f\u043e\u043b\u043d\u0430\u044f \u0442\u0430\u0431\u043b\u0438\u0446\u0430 \u043a\u0432\u0430\u043d\u0442\u043e\u0432 Unsloth<\/h2>\n<table>\n<thead>\n<tr>\n<th>\u041a\u0432\u0430\u043d\u0442<\/th>\n<th>\u0420\u0430\u0437\u043c\u0435\u0440<\/th>\n<th>\u0411\u0438\u0442\/\u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440<\/th>\n<th>\u041c\u0438\u043d\u0438\u043c\u0430\u043b\u044c\u043d\u043e\u0435 \u0436\u0435\u043b\u0435\u0437\u043e<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>UD-IQ1_S<\/td>\n<td>72.5 \u0413\u0411<\/td>\n<td>~1<\/td>\n<td>DGX Spark (\u0441 offloading)<\/td>\n<\/tr>\n<tr>\n<td>UD-IQ1_M<\/td>\n<td>74.5 \u0413\u0411<\/td>\n<td>~1<\/td>\n<td>DGX Spark (\u0441 offloading)<\/td>\n<\/tr>\n<tr>\n<td>UD-Q2_K_XL<\/td>\n<td>78.9 \u0413\u0411<\/td>\n<td>~2<\/td>\n<td>DGX Spark (\u0441 offloading)<\/td>\n<\/tr>\n<tr>\n<td>UD-IQ3_XXS<\/td>\n<td>82 \u0413\u0411<\/td>\n<td>~3<\/td>\n<td>DGX Spark (\u043f\u0440\u0435\u0434\u0435\u043b)<\/td>\n<\/tr>\n<tr>\n<td>UD-Q3_K_XL<\/td>\n<td>90 \u0413\u0411<\/td>\n<td>~3<\/td>\n<td>DGX Spark + NVMe swap<\/td>\n<\/tr>\n<tr>\n<td>UD-IQ4_XS<\/td>\n<td>93.7 \u0413\u0411<\/td>\n<td>~4<\/td>\n<td>2\u00d7 A6000 \/ A100 160G<\/td>\n<\/tr>\n<tr>\n<td>UD-Q4_K_XL<\/td>\n<td>111 \u0413\u0411<\/td>\n<td>~4<\/td>\n<td>A100 160G \/ 2\u00d7 A6000<\/td>\n<\/tr>\n<tr>\n<td>UD-Q5_K_XL<\/td>\n<td>158 \u0413\u0411<\/td>\n<td>~5<\/td>\n<td>A100 80G \u00d7 2<\/td>\n<\/tr>\n<tr>\n<td>UD-Q6_K_XL<\/td>\n<td>169 \u0413\u0411<\/td>\n<td>~6<\/td>\n<td>A100 80G \u00d7 2\u20133<\/td>\n<\/tr>\n<tr>\n<td>Q8_0<\/td>\n<td>188 \u0413\u0411<\/td>\n<td>8<\/td>\n<td>A100 80G \u00d7 3<\/td>\n<\/tr>\n<tr>\n<td>BF16<\/td>\n<td>354 \u0413\u0411<\/td>\n<td>16<\/td>\n<td>A100 80G \u00d7 5+<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<figure><img decoding=\"async\" alt=\"Unsloth-\u043a\u0432\u0430\u043d\u0442\u044b: \u0440\u0430\u0437\u043c\u0435\u0440 vs \u043a\u0430\u0447\u0435\u0441\u0442\u0432\u043e\" src=\"https:\/\/ip-calculator.ru\/blog\/wp-content\/uploads\/2026\/09\/unsloth-quants.png\" \/><figcaption>\u0410\u0432\u0442\u043e\u0440\u0441\u043a\u0430\u044f \u0442\u0430\u0431\u043b\u0438\u0446\u0430: Unsloth Dynamic 3.0 \u043a\u0432\u0430\u043d\u0442\u044b \u0434\u043b\u044f Qwen3.8-Flash-Next. \u0420\u0435\u043a\u043e\u043c\u0435\u043d\u0434\u0430\u0446\u0438\u044f Simon Willison \u0434\u043b\u044f DGX Spark \u2014 UD-Q2_K_XL.<\/figcaption><\/figure>\n<h2>\u041f\u0440\u0430\u043a\u0442\u0438\u043a\u0430: \u043a\u0430\u043a \u0441\u043a\u0430\u0447\u0430\u0442\u044c \u0438 \u0437\u0430\u043f\u0443\u0441\u0442\u0438\u0442\u044c<\/h2>\n<h3>\u0412\u0430\u0440\u0438\u0430\u043d\u0442 1: llama.cpp (\u0440\u0435\u043a\u043e\u043c\u0435\u043d\u0434\u0443\u0435\u0442\u0441\u044f \u0434\u043b\u044f \u043a\u0432\u0430\u043d\u0442\u043e\u0432)<\/h3>\n<pre><code class=\"language-bash\"># \u0423\u0441\u0442\u0430\u043d\u043e\u0432\u043a\u0430 llama.cpp\ncurl -LsSf https:\/\/llama.app\/install.sh | sh\n\n# \u0417\u0430\u043f\u0443\u0441\u043a \u0441 UD-Q4_K_XL (111 \u0413\u0411 \u2014 \u043d\u0443\u0436\u0435\u043d \u0441\u0435\u0440\u0432\u0435\u0440 \u0441 \u0434\u043e\u0441\u0442\u0430\u0442\u043e\u0447\u043d\u044b\u043c VRAM)\nllama serve -hf unsloth\/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL\n\n# \u0418\u043b\u0438 \u0447\u0435\u0440\u0435\u0437 Docker\ndocker model run hf.co\/unsloth\/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL\n<\/code><\/pre>\n<h3>\u0412\u0430\u0440\u0438\u0430\u043d\u0442 2: Ollama<\/h3>\n<pre><code class=\"language-bash\"># \u041f\u0440\u044f\u043c\u043e\u0439 \u0437\u0430\u043f\u0443\u0441\u043a (\u0441\u043a\u0430\u0447\u0430\u0435\u0442 GGUF \u0430\u0432\u0442\u043e\u043c\u0430\u0442\u0438\u0447\u0435\u0441\u043a\u0438)\nollama run hf.co\/unsloth\/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL\n<\/code><\/pre>\n<h3>\u0412\u0430\u0440\u0438\u0430\u043d\u0442 3: vLLM (\u043f\u043e\u043b\u043d\u044b\u0435 \u0432\u0435\u0441\u0430 \u0438\u043b\u0438 GGUF)<\/h3>\n<pre><code class=\"language-bash\">pip install vllm\nvllm serve &quot;Qwen\/Qwen3.8-Flash-Next&quot;\n\n# \u0417\u0430\u043f\u0440\u043e\u0441\ncurl -X POST &quot;http:\/\/localhost:8000\/v1\/chat\/completions&quot; \\\n  -H &quot;Content-Type: application\/json&quot; \\\n  --data '{\n    &quot;model&quot;: &quot;Qwen\/Qwen3.8-Flash-Next&quot;,\n    &quot;messages&quot;: [{&quot;role&quot;: &quot;user&quot;, &quot;content&quot;: &quot;\u041d\u0430\u043f\u0438\u0448\u0438 hello world \u043d\u0430 Python&quot;}]\n  }'\n<\/code><\/pre>\n<h3>\u0412\u0430\u0440\u0438\u0430\u043d\u0442 4: SGLang<\/h3>\n<pre><code class=\"language-bash\">pip install sglang\npython3 -m sglang.launch_server \\\n    --model-path &quot;Qwen\/Qwen3.8-Flash-Next&quot; \\\n    --host 0.0.0.0 --port 30000\n<\/code><\/pre>\n<h3>\u0412\u0430\u0440\u0438\u0430\u043d\u0442 5: LM Studio (\u0432\u0438\u0437\u0443\u0430\u043b\u044c\u043d\u044b\u0439 \u0438\u043d\u0442\u0435\u0440\u0444\u0435\u0439\u0441)<\/h3>\n<p>LM Studio (https:\/\/lmstudio.ai\/) \u2014 \u0434\u0435\u0441\u043a\u0442\u043e\u043f\u043d\u043e\u0435 \u043f\u0440\u0438\u043b\u043e\u0436\u0435\u043d\u0438\u0435 \u0441 GUI. \u041e\u0442\u043a\u0440\u044b\u0432\u0430\u0435\u0442\u0435 Models, \u0438\u0449\u0435\u0442\u0435 <code>unsloth\/Qwen3.8-Flash-Next-GGUF<\/code>, \u0441\u043a\u0430\u0447\u0438\u0432\u0430\u0435\u0442\u0435 \u043a\u0432\u0430\u043d\u0442 \u043d\u0443\u0436\u043d\u043e\u0433\u043e \u0440\u0430\u0437\u043c\u0435\u0440\u0430 \u043e\u0434\u043d\u0438\u043c \u043a\u043b\u0438\u043a\u043e\u043c. \u0417\u0430\u043f\u0443\u0441\u043a\u0430\u0435\u0442\u0435 \u0441\u0435\u0440\u0432\u0435\u0440, \u043f\u043e\u0434\u043a\u043b\u044e\u0447\u0430\u0435\u0442\u0435 \u0430\u0433\u0435\u043d\u0442\u0430 \u043f\u043e OpenAI-\u0441\u043e\u0432\u043c\u0435\u0441\u0442\u0438\u043c\u043e\u043c\u0443 API \u043d\u0430 <code>localhost:1234<\/code>.<\/p>\n<h3>\u041d\u0430\u0441\u0442\u0440\u043e\u0439\u043a\u0430 thinking mode<\/h3>\n<p>\u041f\u043e \u0443\u043c\u043e\u043b\u0447\u0430\u043d\u0438\u044e \u043c\u043e\u0434\u0435\u043b\u044c \u0440\u0430\u0431\u043e\u0442\u0430\u0435\u0442 \u0432 thinking mode \u2014 \u0433\u0435\u043d\u0435\u0440\u0438\u0440\u0443\u0435\u0442 \u00ab\u0440\u0430\u0441\u0441\u0443\u0436\u0434\u0435\u043d\u0438\u044f\u00bb \u043f\u0435\u0440\u0435\u0434 \u043e\u0442\u0432\u0435\u0442\u043e\u043c. \u0414\u043b\u044f \u044d\u043a\u043e\u043d\u043e\u043c\u0438\u0438 \u0442\u043e\u043a\u0435\u043d\u043e\u0432 \u043d\u0430 \u0440\u0443\u0442\u0438\u043d\u043d\u044b\u0445 \u0437\u0430\u0434\u0430\u0447\u0430\u0445:<\/p>\n<pre><code class=\"language-python\"># \u041e\u0442\u043a\u043b\u044e\u0447\u0435\u043d\u0438\u0435 thinking \u0447\u0435\u0440\u0435\u0437 API\nextra_body = {&quot;chat_template_kwargs&quot;: {&quot;enable_thinking&quot;: False}}\n<\/code><\/pre>\n<p>\u0423\u0440\u043e\u0432\u043d\u0438 reasoning effort: <code>xhigh<\/code> (\u043c\u0430\u043a\u0441\u0438\u043c\u0443\u043c), <code>medium<\/code>, <code>low<\/code>. Simon Willison \u0440\u0435\u043a\u043e\u043c\u0435\u043d\u0434\u0443\u0435\u0442 <code>low<\/code> \u0434\u043b\u044f \u0433\u0435\u043d\u0435\u0440\u0430\u0446\u0438\u0438 \u043a\u043e\u0434\u0430 \u2014 \u043d\u0430 DGX Spark \u0441 <code>xhigh<\/code> \u043f\u0440\u043e\u0441\u0442\u0430\u044f SVG-\u0437\u0430\u0434\u0430\u0447\u0430 \u043f\u0440\u0435\u0432\u0440\u0430\u0449\u0430\u043b\u0430\u0441\u044c \u0432 21-\u043c\u0438\u043d\u0443\u0442\u043d\u043e\u0435 \u043e\u0436\u0438\u0434\u0430\u043d\u0438\u0435.<\/p>\n<h2>\u0422\u0440\u0435\u0431\u043e\u0432\u0430\u043d\u0438\u044f \u043f\u043e \u0436\u0435\u043b\u0435\u0437\u0443<\/h2>\n<p>\u041c\u0438\u043d\u0438\u043c\u0430\u043b\u044c\u043d\u044b\u0435 \u0438 \u0440\u0435\u043a\u043e\u043c\u0435\u043d\u0434\u0443\u0435\u043c\u044b\u0435 \u043a\u043e\u043d\u0444\u0438\u0433\u0443\u0440\u0430\u0446\u0438\u0438:<\/p>\n<table>\n<thead>\n<tr>\n<th>\u0421\u0446\u0435\u043d\u0430\u0440\u0438\u0439<\/th>\n<th>\u0416\u0435\u043b\u0435\u0437\u043e<\/th>\n<th>\u041a\u0432\u0430\u043d\u0442<\/th>\n<th>\u041e\u0436\u0438\u0434\u0430\u0435\u043c\u0430\u044f \u0441\u043a\u043e\u0440\u043e\u0441\u0442\u044c<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>\u042d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442<\/td>\n<td>DGX Spark (48 \u0413\u0411)<\/td>\n<td>UD-IQ1_S\/Q2_K_XL (\u0441 offloading)<\/td>\n<td>\u041c\u0435\u0434\u043b\u0435\u043d\u043d\u0430\u044f<\/td>\n<\/tr>\n<tr>\n<td>\u0420\u0430\u0431\u043e\u0447\u0438\u0439 \u043b\u043e\u043a\u0430\u043b\u044c\u043d\u044b\u0439<\/td>\n<td>2\u00d7 A6000 (96 \u0413\u0411)<\/td>\n<td>UD-Q4_K_XL (111 \u0413\u0411)<\/td>\n<td>\u0421\u0440\u0435\u0434\u043d\u044f\u044f<\/td>\n<\/tr>\n<tr>\n<td>\u0421\u0435\u0440\u0432\u0435\u0440<\/td>\n<td>A100 160G<\/td>\n<td>UD-Q4_K_XL (111 \u0413\u0411)<\/td>\n<td>\u0412\u044b\u0441\u043e\u043a\u0430\u044f<\/td>\n<\/tr>\n<tr>\n<td>\u041c\u0430\u043a\u0441\u0438\u043c\u0443\u043c<\/td>\n<td>A100 80G \u00d7 3<\/td>\n<td>Q8_0 (188 \u0413\u0411)<\/td>\n<td>\u041c\u0430\u043a\u0441\u0438\u043c\u0430\u043b\u044c\u043d\u0430\u044f<\/td>\n<\/tr>\n<tr>\n<td>\u041e\u0431\u043b\u0430\u0447\u043d\u044b\u0439 API<\/td>\n<td>\u2014<\/td>\n<td>\u2014<\/td>\n<td>\u0417\u0430\u0432\u0438\u0441\u0438\u0442 \u043e\u0442 \u043f\u0440\u043e\u0432\u0430\u0439\u0434\u0435\u0440\u0430<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>DGX Spark \u2014 \u043a\u043e\u043c\u043f\u0430\u043a\u0442\u043d\u044b\u0439 AI-\u0441\u0435\u0440\u0432\u0435\u0440 \u043e\u0442 NVIDIA, \u043e\u0444\u0438\u0446\u0438\u0430\u043b\u044c\u043d\u043e \u0441\u0442\u043e\u0438\u0442 \u043e\u043a\u043e\u043b\u043e $3 000. 48 \u0413\u0411 unified memory, ARM-\u043f\u0440\u043e\u0446\u0435\u0441\u0441\u043e\u0440. Simon Willison \u0430\u043a\u0442\u0438\u0432\u043d\u043e \u0442\u0435\u0441\u0442\u0438\u0440\u0443\u0435\u0442 \u043d\u0430 \u043d\u0451\u043c \u043b\u043e\u043a\u0430\u043b\u044c\u043d\u044b\u0435 \u043c\u043e\u0434\u0435\u043b\u0438 \u2014 \u044d\u0442\u043e \u0435\u0433\u043e \u043e\u0441\u043d\u043e\u0432\u043d\u0430\u044f \u043f\u043b\u0430\u0442\u0444\u043e\u0440\u043c\u0430 \u0434\u043b\u044f \u044d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442\u043e\u0432 \u0441 \u043e\u0442\u043a\u0440\u044b\u0442\u044b\u043c\u0438 \u0432\u0435\u0441\u0430\u043c\u0438.<\/p>\n<h2>\u0420\u043e\u0441\u0441\u0438\u0439\u0441\u043a\u0438\u0435 \u0440\u0435\u0430\u043b\u0438\u0438<\/h2>\n<h3>\u0414\u043e\u0441\u0442\u0443\u043f\u043d\u043e\u0441\u0442\u044c \u043c\u043e\u0434\u0435\u043b\u0438<\/h3>\n<p><strong>Hugging Face<\/strong> \u2014 \u043e\u0441\u043d\u043e\u0432\u043d\u043e\u0439 \u0438\u0441\u0442\u043e\u0447\u043d\u0438\u043a \u0432\u0435\u0441\u043e\u0432. \u0412 \u0420\u043e\u0441\u0441\u0438\u0438 HF \u0440\u0430\u0431\u043e\u0442\u0430\u0435\u0442 \u0431\u0435\u0437 VPN (https:\/\/huggingface.co), \u043d\u043e \u0441\u043a\u0430\u0447\u0438\u0432\u0430\u043d\u0438\u0435 72\u2013354 \u0413\u0411 \u0432\u0435\u0441\u043e\u0432 \u0442\u0440\u0435\u0431\u0443\u0435\u0442 \u0441\u0442\u0430\u0431\u0438\u043b\u044c\u043d\u043e\u0433\u043e \u0441\u043e\u0435\u0434\u0438\u043d\u0435\u043d\u0438\u044f. \u0417\u0435\u0440\u043a\u0430\u043b\u043e hf-mirror.com (https:\/\/hf-mirror.com) \u0443\u0441\u043a\u043e\u0440\u044f\u0435\u0442 \u0441\u043a\u0430\u0447\u0438\u0432\u0430\u043d\u0438\u0435 \u0447\u0435\u0440\u0435\u0437 \u043a\u0438\u0442\u0430\u0439\u0441\u043a\u0438\u0435 CDN-\u0441\u0435\u0440\u0432\u0435\u0440\u044b Alibaba.<\/p>\n<p><strong>Qwen Cloud<\/strong> (https:\/\/www.qwencloud.com) \u2014 \u043e\u0431\u043b\u0430\u0447\u043d\u044b\u0439 API. \u0414\u043e\u0441\u0442\u0443\u043f\u043d\u043e\u0441\u0442\u044c \u0438\u0437 \u0420\u0424 \u043d\u0435 \u043f\u0440\u043e\u0432\u0435\u0440\u0435\u043d\u0430; Alibaba \u043d\u0435 \u043f\u0443\u0431\u043b\u0438\u043a\u0443\u0435\u0442 \u0441\u043f\u0438\u0441\u043e\u043a \u0437\u0430\u0431\u043b\u043e\u043a\u0438\u0440\u043e\u0432\u0430\u043d\u043d\u044b\u0445 \u0440\u0435\u0433\u0438\u043e\u043d\u043e\u0432.<\/p>\n<p><strong>Unsloth<\/strong> \u2014 \u043a\u0432\u0430\u043d\u0442\u044b \u0434\u043e\u0441\u0442\u0443\u043f\u043d\u044b \u0447\u0435\u0440\u0435\u0437 HF (https:\/\/huggingface.co\/unsloth\/Qwen3.8-Flash-Next-GGUF), \u0441\u043a\u0430\u0447\u0438\u0432\u0430\u043d\u0438\u0435 \u0447\u0435\u0440\u0435\u0437 hf-mirror.com.<\/p>\n<h3>DGX Spark \u0432 \u0420\u0424<\/h3>\n<p>NVIDIA DGX Spark \u043e\u0444\u0438\u0446\u0438\u0430\u043b\u044c\u043d\u043e \u043d\u0435 \u043f\u0440\u043e\u0434\u0430\u0451\u0442\u0441\u044f \u0432 \u0420\u043e\u0441\u0441\u0438\u0438. \u0427\u0435\u0440\u0435\u0437 \u043f\u043e\u0441\u0440\u0435\u0434\u043d\u0438\u043a\u043e\u0432 \u0438 \u043f\u0435\u0440\u0435\u043a\u0443\u043f\u0449\u0438\u043a\u043e\u0432 \u0434\u043e\u0441\u0442\u0443\u043f\u0435\u043d, \u043d\u043e \u0441 \u043d\u0430\u0446\u0435\u043d\u043a\u043e\u0439 30\u201350% \u0438 \u0432\u043e\u043f\u0440\u043e\u0441\u0430\u043c\u0438 \u0433\u0430\u0440\u0430\u043d\u0442\u0438\u0438. \u0410\u043b\u044c\u0442\u0435\u0440\u043d\u0430\u0442\u0438\u0432\u044b \u0434\u043b\u044f \u043b\u043e\u043a\u0430\u043b\u044c\u043d\u043e\u0433\u043e \u0437\u0430\u043f\u0443\u0441\u043a\u0430:<\/p>\n<ul>\n<li><strong>\u0420\u0430\u0431\u043e\u0447\u0430\u044f \u0441\u0442\u0430\u043d\u0446\u0438\u044f \u0441 2\u00d7 NVIDIA RTX 4090 (48 \u0413\u0411 VRAM)<\/strong> \u2014 \u0445\u0432\u0430\u0442\u0438\u0442 \u0434\u043b\u044f UD-IQ1_S (72.5 \u0413\u0411 \u0441 offloading). \u0421\u0442\u043e\u0438\u043c\u043e\u0441\u0442\u044c 400\u2013500 \u0442\u044b\u0441. \u0440\u0443\u0431.<\/li>\n<li><strong>\u0421\u0435\u0440\u0432\u0435\u0440 \u0441 NVIDIA A6000 (48 \u0413\u0411) \u00d7 2<\/strong> \u2014 \u0434\u043b\u044f UD-Q2_K_XL (78.9 \u0413\u0411). \u0421\u0442\u043e\u0438\u043c\u043e\u0441\u0442\u044c \u043e\u0442 800 \u0442\u044b\u0441. \u0440\u0443\u0431.<\/li>\n<li><strong>\u0410\u0440\u0435\u043d\u0434\u0430 GPU<\/strong> \u2014 vast.ai, RunPod (\u043f\u0440\u0438\u043d\u0438\u043c\u0430\u044e\u0442 \u043a\u0440\u0438\u043f\u0442\u0443). A100 80G \u2014 \u043e\u0442 $1\/\u0447\u0430\u0441.<\/li>\n<\/ul>\n<h3>\u041e\u043f\u043b\u0430\u0442\u0430<\/h3>\n<p>\u0420\u043e\u0441\u0441\u0438\u0439\u0441\u043a\u0438\u0435 Visa\/Mastercard \u043d\u0435 \u0440\u0430\u0431\u043e\u0442\u0430\u044e\u0442 \u0441 \u0437\u0430\u0440\u0443\u0431\u0435\u0436\u043d\u044b\u043c\u0438 \u0441\u0435\u0440\u0432\u0438\u0441\u0430\u043c\u0438. \u0412\u0430\u0440\u0438\u0430\u043d\u0442\u044b:<br \/>\n&#8212; <strong>\u041b\u043e\u043a\u0430\u043b\u044c\u043d\u044b\u0439 \u0437\u0430\u043f\u0443\u0441\u043a<\/strong> \u2014 \u0431\u0435\u0441\u043f\u043b\u0430\u0442\u043d\u043e, \u0431\u0435\u0437 \u043e\u043f\u043b\u0430\u0442\u044b. \u042d\u0442\u043e \u0433\u043b\u0430\u0432\u043d\u043e\u0435 \u043f\u0440\u0435\u0438\u043c\u0443\u0449\u0435\u0441\u0442\u0432\u043e open-weight \u043c\u043e\u0434\u0435\u043b\u0435\u0439.<br \/>\n&#8212; <strong>\u0410\u0440\u0435\u043d\u0434\u0430 GPU<\/strong> \u2014 \u0447\u0435\u0440\u0435\u0437 \u043a\u0440\u0438\u043f\u0442\u043e\u0432\u0430\u043b\u044e\u0442\u0443 (BTC\/USDT) \u0438\u043b\u0438 UnionPay.<br \/>\n&#8212; <strong>\u041f\u043e\u0441\u0440\u0435\u0434\u043d\u0438\u043a\u0438<\/strong> \u2014 AITUNNEL (https:\/\/aitunnel.ru\/) \u043f\u0440\u0438\u043d\u0438\u043c\u0430\u0435\u0442 \u0440\u0443\u0431\u043b\u0438 \u0437\u0430 \u0447\u0430\u0441\u0442\u044c \u0441\u0435\u0440\u0432\u0438\u0441\u043e\u0432.<\/p>\n<p>\u0422\u0435\u043c\u0430 \u0437\u0430\u043f\u0443\u0441\u043a\u0430 Qwen3.8-Flash-Next \u0438 DGX Spark \u0432 \u0440\u0443\u043d\u0435\u0442\u0435 \u043f\u0440\u0430\u043a\u0442\u0438\u0447\u0435\u0441\u043a\u0438 \u043d\u0435 \u043e\u0441\u0432\u0435\u0449\u0435\u043d\u0430. Simon Willison \u2014 \u043e\u0441\u043d\u043e\u0432\u043d\u043e\u0439 \u0438\u0441\u0442\u043e\u0447\u043d\u0438\u043a \u043f\u0440\u0430\u043a\u0442\u0438\u0447\u0435\u0441\u043a\u043e\u0433\u043e \u043e\u043f\u044b\u0442\u0430.<\/p>\n<h2>\u0427\u0435\u0441\u0442\u043d\u044b\u0435 \u043e\u0433\u0440\u0430\u043d\u0438\u0447\u0435\u043d\u0438\u044f<\/h2>\n<ul>\n<li><strong>\u0420\u0430\u0437\u043c\u0435\u0440 \u0432\u0435\u0441\u043e\u0432.<\/strong> \u0414\u0430\u0436\u0435 \u0441\u0430\u043c\u044b\u0439 \u043a\u043e\u043c\u043f\u0430\u043a\u0442\u043d\u044b\u0439 \u043a\u0432\u0430\u043d\u0442 (IQ1_S) \u0432\u0435\u0441\u0438\u0442 72.5 \u0413\u0411. \u041d\u0430 \u043d\u043e\u0443\u0442\u0431\u0443\u043a\u0435 \u0441 16 \u0413\u0411 RAM \u2014 \u043d\u0435 \u0437\u0430\u043f\u0443\u0441\u0442\u0438\u0442\u0441\u044f. \u042d\u0442\u043e \u0441\u0435\u0440\u0432\u0435\u0440\u043d\u0430\u044f \u043c\u043e\u0434\u0435\u043b\u044c.<\/li>\n<li><strong>Offloading = \u043c\u0435\u0434\u043b\u0435\u043d\u043d\u043e.<\/strong> DGX Spark (48 \u0413\u0411) \u043d\u0435 \u0432\u043c\u0435\u0449\u0430\u0435\u0442 \u043a\u0432\u0430\u043d\u0442\u044b \u0446\u0435\u043b\u0438\u043a\u043e\u043c. \u0412\u044b\u0433\u0440\u0443\u0437\u043a\u0430 \u043d\u0430 NVMe \u0437\u0430\u043c\u0435\u0434\u043b\u044f\u0435\u0442 \u0433\u0435\u043d\u0435\u0440\u0430\u0446\u0438\u044e \u0432 3\u20135 \u0440\u0430\u0437 \u043f\u043e \u0441\u0440\u0430\u0432\u043d\u0435\u043d\u0438\u044e \u0441 \u043f\u043e\u043b\u043d\u044b\u043c VRAM.<\/li>\n<li><strong>1-\u0431\u0438\u0442\u043d\u043e\u0435 \u043a\u0432\u0430\u043d\u0442\u043e\u0432\u0430\u043d\u0438\u0435 \u0434\u0435\u0433\u0440\u0430\u0434\u0438\u0440\u0443\u0435\u0442.<\/strong> UD-IQ1_S (72.5 \u0413\u0411) \u2014 \u044d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442\u0430\u043b\u044c\u043d\u044b\u0439. \u0414\u043b\u044f \u0440\u0435\u0430\u043b\u044c\u043d\u043e\u0439 \u0440\u0430\u0431\u043e\u0442\u044b \u043d\u0443\u0436\u0435\u043d \u043c\u0438\u043d\u0438\u043c\u0443\u043c Q2_K_XL (78.9 \u0413\u0411), \u043b\u0443\u0447\u0448\u0435 Q3_K_XL (90 \u0413\u0411) \u0438\u043b\u0438 Q4_K_XL (111 \u0413\u0411).<\/li>\n<li><strong>Reasoning \u043f\u043e \u0443\u043c\u043e\u043b\u0447\u0430\u043d\u0438\u044e.<\/strong> \u041c\u043e\u0434\u0435\u043b\u044c \u0442\u0440\u0430\u0442\u0438\u0442 \u0434\u0435\u0441\u044f\u0442\u043a\u0438 \u0442\u044b\u0441\u044f\u0447 \u0442\u043e\u043a\u0435\u043d\u043e\u0432 \u043d\u0430 \u00ab\u0440\u0430\u0441\u0441\u0443\u0436\u0434\u0435\u043d\u0438\u044f\u00bb \u0434\u0430\u0436\u0435 \u0434\u043b\u044f \u043f\u0440\u043e\u0441\u0442\u044b\u0445 \u0437\u0430\u0434\u0430\u0447. \u0411\u0435\u0437 \u043e\u0442\u043a\u043b\u044e\u0447\u0435\u043d\u0438\u044f thinking \u0440\u0435\u0436\u0438\u043c\u0430 \u0441\u043a\u043e\u0440\u043e\u0441\u0442\u044c \u043f\u0430\u0434\u0430\u0435\u0442 \u0432 10 \u0440\u0430\u0437.<\/li>\n<li><strong>\u0410\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u0430 \u044d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442\u0430\u043b\u044c\u043d\u0430\u044f.<\/strong> <code>qwen4exp<\/code> \u2014 \u043f\u0440\u0435\u0432\u044c\u044e. \u0424\u043e\u0440\u043c\u0430\u0442, \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u044b \u0438 \u043f\u043e\u0432\u0435\u0434\u0435\u043d\u0438\u0435 \u043c\u043e\u0433\u0443\u0442 \u0438\u0437\u043c\u0435\u043d\u0438\u0442\u044c\u0441\u044f \u0432 \u0440\u0435\u043b\u0438\u0437\u0435 Qwen4.<\/li>\n<li><strong>\u041b\u0438\u0446\u0435\u043d\u0437\u0438\u044f qwen-community-1.0<\/strong> \u2014 \u043e\u0442\u043a\u0440\u044b\u0442\u0430\u044f, \u043d\u043e \u043d\u0435 Apache 2.0. \u041f\u0440\u043e\u0432\u0435\u0440\u044c\u0442\u0435 \u0443\u0441\u043b\u043e\u0432\u0438\u044f \u0434\u043b\u044f \u043a\u043e\u043c\u043c\u0435\u0440\u0447\u0435\u0441\u043a\u043e\u0433\u043e \u0438\u0441\u043f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u043d\u0438\u044f.<\/li>\n<li><strong>\u041c\u0443\u043b\u044c\u0442\u0438\u043c\u043e\u0434\u0430\u043b\u044c\u043d\u043e\u0441\u0442\u044c \u043e\u0433\u0440\u0430\u043d\u0438\u0447\u0435\u043d\u0430.<\/strong> Vision Encoder \u0435\u0441\u0442\u044c, \u043d\u043e \u043d\u0430 \u043f\u0440\u0430\u043a\u0442\u0438\u043a\u0435 \u043d\u0430 DGX Spark \u0441 \u043a\u0432\u0430\u043d\u0442\u043e\u0432\u0430\u043d\u043d\u043e\u0439 \u043c\u043e\u0434\u0435\u043b\u044c\u044e \u043e\u0431\u0440\u0430\u0431\u043e\u0442\u043a\u0430 \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 \u0442\u0440\u0435\u0431\u0443\u0435\u0442 \u0437\u043d\u0430\u0447\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u0445 \u0434\u043e\u043f\u043e\u043b\u043d\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u0445 \u0440\u0435\u0441\u0443\u0440\u0441\u043e\u0432.<\/li>\n<\/ul>\n<h2>\u0412\u044b\u0432\u043e\u0434<\/h2>\n<p>Qwen3.8-Flash-Next \u0434\u0435\u043c\u043e\u043d\u0441\u0442\u0440\u0438\u0440\u0443\u0435\u0442 \u0442\u0440\u0435\u043d\u0434, \u043a\u043e\u0442\u043e\u0440\u044b\u0439 \u043e\u043f\u0440\u0435\u0434\u0435\u043b\u044f\u0435\u0442 2026 \u0433\u043e\u0434: \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u043d\u0430\u044f \u044d\u0444\u0444\u0435\u043a\u0442\u0438\u0432\u043d\u043e\u0441\u0442\u044c \u0432\u0430\u0436\u043d\u0435\u0435 \u0433\u0440\u0443\u0431\u043e\u0433\u043e \u043c\u0430\u0441\u0448\u0442\u0430\u0431\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u044f. 512 \u044d\u043a\u0441\u043f\u0435\u0440\u0442\u043e\u0432 \u0441 11 \u0430\u043a\u0442\u0438\u0432\u043d\u044b\u043c\u0438 \u043d\u0430 \u0448\u0430\u0433, N-gram Embedding \u0434\u043b\u044f \u043c\u0430\u0441\u0448\u0442\u0430\u0431\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u044f \u0447\u0435\u0440\u0435\u0437 \u043f\u0430\u043c\u044f\u0442\u044c, Sparse Attention \u0434\u043b\u044f \u0434\u043b\u0438\u043d\u043d\u043e\u0433\u043e \u043a\u043e\u043d\u0442\u0435\u043a\u0441\u0442\u0430 \u2014 \u0432\u0441\u0451 \u044d\u0442\u043e \u043f\u043e\u0437\u0432\u043e\u043b\u044f\u0435\u0442 6 \u043c\u043b\u0440\u0434 \u0430\u043a\u0442\u0438\u0432\u043d\u044b\u0445 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432 \u043a\u043e\u043d\u043a\u0443\u0440\u0438\u0440\u043e\u0432\u0430\u0442\u044c \u0441 400+ \u043c\u043b\u0440\u0434 \u0444\u043b\u0430\u0433\u043c\u0430\u043d\u043e\u0432 \u043d\u0430 \u043a\u043e\u0434\u0438\u043d\u0433\u043e\u0432\u044b\u0445 \u0437\u0430\u0434\u0430\u0447\u0430\u0445.<\/p>\n<p>\u0414\u043b\u044f \u043f\u0440\u0430\u043a\u0442\u0438\u043a\u0438: \u043c\u043e\u0434\u0435\u043b\u044c \u043f\u043e\u0434\u0445\u043e\u0434\u0438\u0442 \u0442\u0435\u043c, \u0443 \u043a\u043e\u0433\u043e \u0435\u0441\u0442\u044c \u0441\u0435\u0440\u0432\u0435\u0440 \u0441 96+ \u0413\u0411 VRAM (\u0438\u043b\u0438 DGX Spark \u0434\u043b\u044f \u044d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442\u043e\u0432). Qwen 3.8 27B \u0432\u0441\u0451 \u0435\u0449\u0451 \u043f\u0440\u043e\u0449\u0435 \u0438 \u0434\u043e\u0441\u0442\u0443\u043f\u043d\u0435\u0435 \u2014 17 \u0413\u0411 \u043d\u0430 \u043d\u043e\u0443\u0442\u0431\u0443\u043a\u0435, Apache 2.0, 15\u201330 \u0442\u043e\u043a\/\u0441. Flash-Next \u2014 \u0441\u043b\u0435\u0434\u0443\u044e\u0449\u0438\u0439 \u0448\u0430\u0433: MoE-\u044d\u0444\u0444\u0435\u043a\u0442\u0438\u0432\u043d\u043e\u0441\u0442\u044c, \u043c\u0443\u043b\u044c\u0442\u0438\u043c\u043e\u0434\u0430\u043b\u044c\u043d\u043e\u0441\u0442\u044c, \u043f\u0440\u0435\u0432\u044c\u044e \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u044b \u0431\u0443\u0434\u0443\u0449\u0435\u0433\u043e.<\/p>\n<p>\u0414\u043b\u044f \u0420\u043e\u0441\u0441\u0438\u0438 \u043b\u043e\u043a\u0430\u043b\u044c\u043d\u044b\u0435 open-weight \u043c\u043e\u0434\u0435\u043b\u0438 \u2014 \u0435\u0434\u0438\u043d\u0441\u0442\u0432\u0435\u043d\u043d\u044b\u0439 \u043f\u0443\u0442\u044c \u0440\u0430\u0431\u043e\u0442\u0430\u0442\u044c \u0441 \u0441\u043e\u0432\u0440\u0435\u043c\u0435\u043d\u043d\u044b\u043c\u0438 LLM \u0431\u0435\u0437 VPN \u0438 \u0437\u0430\u0440\u0443\u0431\u0435\u0436\u043d\u044b\u0445 \u043a\u0430\u0440\u0442. HF \u0434\u043e\u0441\u0442\u0443\u043f\u0435\u043d, \u043a\u0432\u0430\u043d\u0442\u044b \u043e\u0442 Unsloth \u0441\u043a\u0430\u0447\u0438\u0432\u0430\u044e\u0442\u0441\u044f \u0447\u0435\u0440\u0435\u0437 hf-mirror.com, \u0430 CUDA-\u0441\u043e\u0432\u043c\u0435\u0441\u0442\u0438\u043c\u043e\u0435 \u0436\u0435\u043b\u0435\u0437\u043e \u043c\u043e\u0436\u043d\u043e \u0441\u043e\u0431\u0440\u0430\u0442\u044c \u0438\u043b\u0438 \u0430\u0440\u0435\u043d\u0434\u043e\u0432\u0430\u0442\u044c \u0437\u0430 \u043a\u0440\u0438\u043f\u0442\u0443.<\/p>\n<p>\u0414\u0430\u043b\u044c\u0448\u0435 \u043f\u043e \u0442\u0435\u043c\u0435: <a href=\"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/malenkie-modeli-dlya-kodinga\/\">\u043c\u0430\u043b\u0435\u043d\u044c\u043a\u0438\u0435 \u0431\u044b\u0441\u0442\u0440\u044b\u0435 \u043c\u043e\u0434\u0435\u043b\u0438 \u0434\u043b\u044f \u043a\u043e\u0434\u0438\u043d\u0433\u0430: gpt-5.6-luna \u0438 \u043a\u043e\u043c\u043f\u0430\u043d\u0438\u044f<\/a>, <a href=\"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/lokalnye-nejroseti-dlya-kodinga-qwen3-coder-ollama\/\">\u043b\u043e\u043a\u0430\u043b\u044c\u043d\u044b\u0435 \u043d\u0435\u0439\u0440\u043e\u0441\u0435\u0442\u0438 \u0434\u043b\u044f \u043a\u043e\u0434\u0438\u043d\u0433\u0430: Qwen3-Coder \u0447\u0435\u0440\u0435\u0437 Ollama<\/a>, <a href=\"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/modeli-dlya-kodinga-2026\/\">\u043c\u043e\u0434\u0435\u043b\u0438 \u0434\u043b\u044f \u043a\u043e\u0434\u0438\u043d\u0433\u0430 2026: \u0441\u0440\u0430\u0432\u043d\u0435\u043d\u0438\u0435<\/a>, <a href=\"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/chto-takoe-agentnyj-koding\/\">\u0447\u0442\u043e \u0442\u0430\u043a\u043e\u0435 \u0430\u0433\u0435\u043d\u0442\u043d\u044b\u0439 \u043a\u043e\u0434\u0438\u043d\u0433<\/a>.<\/p>\n<div class=\"fpm_end\"><\/div>\n","protected":false},"excerpt":{"rendered":"<p>26 \u0430\u0432\u0433\u0443\u0441\u0442\u0430 2026 \u0433\u043e\u0434\u0430 Alibaba \u0432\u044b\u043f\u0443\u0441\u0442\u0438\u043b\u0430 Qwen3.8-Flash-Next \u2014 \u044d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442\u0430\u043b\u044c\u043d\u0443\u044e \u043c\u043e\u0434\u0435\u043b\u044c \u0441 \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u043e\u0439, \u043a\u043e\u0442\u043e\u0440\u0443\u044e \u043a\u043e\u043c\u0430\u043d\u0434\u0430 \u043d\u0430\u0437\u044b\u0432\u0430\u0435\u0442 \u00ab\u043f\u0440\u0435\u0432\u044c\u044e Qwen4\u00bb. 125 \u043c\u043b\u0440\u0434 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432, \u043d\u043e \u043d\u0430 \u043a\u0430\u0436\u0434\u043e\u043c \u0448\u0430\u0433\u0435 \u0430\u043a\u0442\u0438\u0432\u043d\u044b \u0442\u043e\u043b\u044c\u043a\u043e 6 \u043c\u043b\u0440\u0434. Simon Willison (\u0430\u0432\u0442\u043e\u0440 \u0431\u0438\u0431\u043b\u0438\u043e\u0442\u0435\u043a\u0438 Datasette \u0438 \u043e\u0434\u0438\u043d \u0438\u0437 \u0441\u0430\u043c\u044b\u0445 \u0430\u043a\u0442\u0438\u0432\u043d\u044b\u0445 \u043e\u0431\u043e\u0437\u0440\u0435\u0432\u0430\u0442\u0435\u043b\u0435\u0439 LLM) \u0441\u043a\u0430\u0447\u0430\u043b \u043a\u0432<\/p>\n","protected":false},"author":5,"featured_media":47843,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"neve_meta_sidebar":"","neve_meta_container":"","neve_meta_enable_content_width":"","neve_meta_content_width":0,"neve_meta_title_alignment":"","neve_meta_author_avatar":"","neve_post_elements_order":"","neve_meta_disable_header":"","neve_meta_disable_footer":"","neve_meta_disable_title":"","footnotes":""},"categories":[262],"tags":[],"class_list":["post-47841","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-artificial-intelligence"],"aioseo_notices":[],"aioseo_head":"\n\t\t<!-- All in One SEO 5.0.5 - aioseo.com -->\n\t<meta name=\"description\" content=\"26 \u0430\u0432\u0433\u0443\u0441\u0442\u0430 2026 \u0433\u043e\u0434\u0430 Alibaba \u0432\u044b\u043f\u0443\u0441\u0442\u0438\u043b\u0430 Qwen3.8-Flash-Next \u2014 \u044d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442\u0430\u043b\u044c\u043d\u0443\u044e \u043c\u043e\u0434\u0435\u043b\u044c \u0441 \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u043e\u0439, \u043a\u043e\u0442\u043e\u0440\u0443\u044e \u043a\u043e\u043c\u0430\u043d\u0434\u0430 \u043d\u0430\u0437\u044b\u0432\u0430\u0435\u0442 \u00ab\u043f\u0440\u0435\u0432\u044c\u044e Qwen4\u00bb. 125 \u043c\u043b\u0440\u0434 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432, \u043d\u043e \u043d\u0430 \u043a\u0430\u0436\u0434\u043e\u043c \u0448\u0430\u0433\u0435 \u0430\u043a\u0442\u0438\u0432\u043d\u044b \u0442\u043e\u043b\u044c\u043a\u043e 6 \u043c\u043b\" \/>\n\t<meta name=\"robots\" content=\"max-image-preview:large\" \/>\n\t<meta name=\"author\" content=\"\u0410\u043d\u0430\u0442\u043e\u043b\u0438\u0439\"\/>\n\t<meta name=\"yandex-verification\" content=\"bf0ffc608a86e312\" \/>\n\t<link rel=\"canonical\" href=\"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/\" \/>\n\t\t<meta name=\"generator\" content=\"All in One SEO (AIOSEO) 5.0.5\" \/>\n\t\t<script type=\"application\/ld+json\" class=\"aioseo-schema\">\n\t\t\t{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#article\",\"name\":\"\\u041a\\u0435\\u0439\\u0441: Qwen3.8-Flash-Next \\u2014 MoE 125B\\\/6B \\u043d\\u0430 DGX Spark\",\"headline\":\"\\u041a\\u0435\\u0439\\u0441: Qwen3.8-Flash-Next \\u2014 MoE 125B\\\/6B \\u043d\\u0430 DGX Spark\",\"author\":{\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/author\\\/anatolyi\\\/#author\"},\"publisher\":{\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/#person\"},\"image\":{\"@type\":\"ImageObject\",\"url\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/wp-content\\\/uploads\\\/2026\\\/08\\\/kejs-qwen38-flash-next-moe-617.jpg\",\"width\":2400,\"height\":1800},\"datePublished\":\"2026-09-05T13:00:00+03:00\",\"dateModified\":\"2026-08-30T03:35:53+03:00\",\"inLanguage\":\"ru-RU\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#webpage\"},\"isPartOf\":{\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#webpage\"},\"articleSection\":\"\\u041f\\u0440\\u043e \\u0438\\u0441\\u043a\\u0443\\u0441\\u0441\\u0442\\u0432\\u0435\\u043d\\u043d\\u044b\\u0439 \\u0438\\u043d\\u0442\\u0435\\u043b\\u043b\\u0435\\u043a\\u0442\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#breadcrumblist\",\"itemListElement\":[{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/#listItem\",\"position\":1,\"name\":\"\\u0421\\u0435\\u0442\\u044c \\u0431\\u0435\\u0437 \\u043f\\u0440\\u043e\\u0431\\u043b\\u0435\\u043c\",\"item\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/\",\"nextItem\":{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/#listItem\",\"name\":\"\\u041f\\u0440\\u043e \\u0438\\u0441\\u043a\\u0443\\u0441\\u0441\\u0442\\u0432\\u0435\\u043d\\u043d\\u044b\\u0439 \\u0438\\u043d\\u0442\\u0435\\u043b\\u043b\\u0435\\u043a\\u0442\"}},{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/#listItem\",\"position\":2,\"name\":\"\\u041f\\u0440\\u043e \\u0438\\u0441\\u043a\\u0443\\u0441\\u0441\\u0442\\u0432\\u0435\\u043d\\u043d\\u044b\\u0439 \\u0438\\u043d\\u0442\\u0435\\u043b\\u043b\\u0435\\u043a\\u0442\",\"item\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/\",\"nextItem\":{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#listItem\",\"name\":\"\\u041a\\u0435\\u0439\\u0441: Qwen3.8-Flash-Next \\u2014 MoE 125B\\\/6B \\u043d\\u0430 DGX Spark\"},\"previousItem\":{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/#listItem\",\"name\":\"\\u0421\\u0435\\u0442\\u044c \\u0431\\u0435\\u0437 \\u043f\\u0440\\u043e\\u0431\\u043b\\u0435\\u043c\"}},{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#listItem\",\"position\":3,\"name\":\"\\u041a\\u0435\\u0439\\u0441: Qwen3.8-Flash-Next \\u2014 MoE 125B\\\/6B \\u043d\\u0430 DGX Spark\",\"previousItem\":{\"@type\":\"ListItem\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/#listItem\",\"name\":\"\\u041f\\u0440\\u043e \\u0438\\u0441\\u043a\\u0443\\u0441\\u0441\\u0442\\u0432\\u0435\\u043d\\u043d\\u044b\\u0439 \\u0438\\u043d\\u0442\\u0435\\u043b\\u043b\\u0435\\u043a\\u0442\"}}]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/#person\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#personImage\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/7ddce1edb52479259959fc679514727cd0d7602b1459163bf6e25a3fc74e7e41?s=96&d=mm&r=g\",\"width\":96,\"height\":96,\"caption\":\"admin\"}},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/author\\\/anatolyi\\\/#author\",\"url\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/author\\\/anatolyi\\\/\",\"name\":\"\\u0410\\u043d\\u0430\\u0442\\u043e\\u043b\\u0438\\u0439\",\"image\":{\"@type\":\"ImageObject\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#authorImage\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/e2d3f7f62377c99d9cfe8b137296f8149892a40f3db52265781ca37e0a55dac6?s=96&d=mm&r=g\",\"width\":96,\"height\":96,\"caption\":\"\\u0410\\u043d\\u0430\\u0442\\u043e\\u043b\\u0438\\u0439\"}},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#webpage\",\"url\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/\",\"name\":\"\\u041a\\u0435\\u0439\\u0441: Qwen3.8-Flash-Next \\u2014 MoE 125B\\\/6B \\u043d\\u0430 DGX Spark\",\"description\":\"26 \\u0430\\u0432\\u0433\\u0443\\u0441\\u0442\\u0430 2026 \\u0433\\u043e\\u0434\\u0430 Alibaba \\u0432\\u044b\\u043f\\u0443\\u0441\\u0442\\u0438\\u043b\\u0430 Qwen3.8-Flash-Next \\u2014 \\u044d\\u043a\\u0441\\u043f\\u0435\\u0440\\u0438\\u043c\\u0435\\u043d\\u0442\\u0430\\u043b\\u044c\\u043d\\u0443\\u044e \\u043c\\u043e\\u0434\\u0435\\u043b\\u044c \\u0441 \\u0430\\u0440\\u0445\\u0438\\u0442\\u0435\\u043a\\u0442\\u0443\\u0440\\u043e\\u0439, \\u043a\\u043e\\u0442\\u043e\\u0440\\u0443\\u044e \\u043a\\u043e\\u043c\\u0430\\u043d\\u0434\\u0430 \\u043d\\u0430\\u0437\\u044b\\u0432\\u0430\\u0435\\u0442 \\u00ab\\u043f\\u0440\\u0435\\u0432\\u044c\\u044e Qwen4\\u00bb. 125 \\u043c\\u043b\\u0440\\u0434 \\u043f\\u0430\\u0440\\u0430\\u043c\\u0435\\u0442\\u0440\\u043e\\u0432, \\u043d\\u043e \\u043d\\u0430 \\u043a\\u0430\\u0436\\u0434\\u043e\\u043c \\u0448\\u0430\\u0433\\u0435 \\u0430\\u043a\\u0442\\u0438\\u0432\\u043d\\u044b \\u0442\\u043e\\u043b\\u044c\\u043a\\u043e 6 \\u043c\\u043b\",\"inLanguage\":\"ru-RU\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/#website\"},\"breadcrumb\":{\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#breadcrumblist\"},\"author\":{\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/author\\\/anatolyi\\\/#author\"},\"creator\":{\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/author\\\/anatolyi\\\/#author\"},\"image\":{\"@type\":\"ImageObject\",\"url\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/wp-content\\\/uploads\\\/2026\\\/08\\\/kejs-qwen38-flash-next-moe-617.jpg\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#mainImage\",\"width\":2400,\"height\":1800},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/artificial-intelligence\\\/kejs-qwen38-flash-next-moe\\\/#mainImage\"},\"datePublished\":\"2026-09-05T13:00:00+03:00\",\"dateModified\":\"2026-08-30T03:35:53+03:00\"},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/#website\",\"url\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/\",\"name\":\"\\u0421\\u0435\\u0442\\u044c \\u0431\\u0435\\u0437 \\u043f\\u0440\\u043e\\u0431\\u043b\\u0435\\u043c\",\"description\":\"\\u041d\\u0430\\u0441\\u0442\\u0440\\u043e\\u0439\\u043a\\u0430 \\u0438 \\u043e\\u0431\\u0441\\u043b\\u0443\\u0436\\u0438\\u0432\\u0430\\u043d\\u0438\\u0435 \\u0441\\u0435\\u0442\\u0438\",\"inLanguage\":\"ru-RU\",\"publisher\":{\"@id\":\"https:\\\/\\\/ip-calculator.ru\\\/blog\\\/#person\"}}]}\n\t\t<\/script>\n\t\t<!-- All in One SEO -->\n\n","aioseo_head_json":{"title":"\u041a\u0435\u0439\u0441: Qwen3.8-Flash-Next \u2014 MoE 125B\/6B \u043d\u0430 DGX Spark","description":"26 \u0430\u0432\u0433\u0443\u0441\u0442\u0430 2026 \u0433\u043e\u0434\u0430 Alibaba \u0432\u044b\u043f\u0443\u0441\u0442\u0438\u043b\u0430 Qwen3.8-Flash-Next \u2014 \u044d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442\u0430\u043b\u044c\u043d\u0443\u044e \u043c\u043e\u0434\u0435\u043b\u044c \u0441 \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u043e\u0439, \u043a\u043e\u0442\u043e\u0440\u0443\u044e \u043a\u043e\u043c\u0430\u043d\u0434\u0430 \u043d\u0430\u0437\u044b\u0432\u0430\u0435\u0442 \u00ab\u043f\u0440\u0435\u0432\u044c\u044e Qwen4\u00bb. 125 \u043c\u043b\u0440\u0434 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432, \u043d\u043e \u043d\u0430 \u043a\u0430\u0436\u0434\u043e\u043c \u0448\u0430\u0433\u0435 \u0430\u043a\u0442\u0438\u0432\u043d\u044b \u0442\u043e\u043b\u044c\u043a\u043e 6 \u043c\u043b","canonical_url":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/","robots":"max-image-preview:large","keywords":"","webmasterTools":{"yandex-verification":"bf0ffc608a86e312","miscellaneous":""},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#article","name":"\u041a\u0435\u0439\u0441: Qwen3.8-Flash-Next \u2014 MoE 125B\/6B \u043d\u0430 DGX Spark","headline":"\u041a\u0435\u0439\u0441: Qwen3.8-Flash-Next \u2014 MoE 125B\/6B \u043d\u0430 DGX Spark","author":{"@id":"https:\/\/ip-calculator.ru\/blog\/author\/anatolyi\/#author"},"publisher":{"@id":"https:\/\/ip-calculator.ru\/blog\/#person"},"image":{"@type":"ImageObject","url":"https:\/\/ip-calculator.ru\/blog\/wp-content\/uploads\/2026\/08\/kejs-qwen38-flash-next-moe-617.jpg","width":2400,"height":1800},"datePublished":"2026-09-05T13:00:00+03:00","dateModified":"2026-08-30T03:35:53+03:00","inLanguage":"ru-RU","mainEntityOfPage":{"@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#webpage"},"isPartOf":{"@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#webpage"},"articleSection":"\u041f\u0440\u043e \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u044b\u0439 \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442"},{"@type":"BreadcrumbList","@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#breadcrumblist","itemListElement":[{"@type":"ListItem","@id":"https:\/\/ip-calculator.ru\/blog\/#listItem","position":1,"name":"\u0421\u0435\u0442\u044c \u0431\u0435\u0437 \u043f\u0440\u043e\u0431\u043b\u0435\u043c","item":"https:\/\/ip-calculator.ru\/blog\/","nextItem":{"@type":"ListItem","@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/#listItem","name":"\u041f\u0440\u043e \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u044b\u0439 \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442"}},{"@type":"ListItem","@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/#listItem","position":2,"name":"\u041f\u0440\u043e \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u044b\u0439 \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442","item":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/","nextItem":{"@type":"ListItem","@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#listItem","name":"\u041a\u0435\u0439\u0441: Qwen3.8-Flash-Next \u2014 MoE 125B\/6B \u043d\u0430 DGX Spark"},"previousItem":{"@type":"ListItem","@id":"https:\/\/ip-calculator.ru\/blog\/#listItem","name":"\u0421\u0435\u0442\u044c \u0431\u0435\u0437 \u043f\u0440\u043e\u0431\u043b\u0435\u043c"}},{"@type":"ListItem","@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#listItem","position":3,"name":"\u041a\u0435\u0439\u0441: Qwen3.8-Flash-Next \u2014 MoE 125B\/6B \u043d\u0430 DGX Spark","previousItem":{"@type":"ListItem","@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/#listItem","name":"\u041f\u0440\u043e \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u044b\u0439 \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442"}}]},{"@type":"Person","@id":"https:\/\/ip-calculator.ru\/blog\/#person","name":"admin","image":{"@type":"ImageObject","@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#personImage","url":"https:\/\/secure.gravatar.com\/avatar\/7ddce1edb52479259959fc679514727cd0d7602b1459163bf6e25a3fc74e7e41?s=96&d=mm&r=g","width":96,"height":96,"caption":"admin"}},{"@type":"Person","@id":"https:\/\/ip-calculator.ru\/blog\/author\/anatolyi\/#author","url":"https:\/\/ip-calculator.ru\/blog\/author\/anatolyi\/","name":"\u0410\u043d\u0430\u0442\u043e\u043b\u0438\u0439","image":{"@type":"ImageObject","@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#authorImage","url":"https:\/\/secure.gravatar.com\/avatar\/e2d3f7f62377c99d9cfe8b137296f8149892a40f3db52265781ca37e0a55dac6?s=96&d=mm&r=g","width":96,"height":96,"caption":"\u0410\u043d\u0430\u0442\u043e\u043b\u0438\u0439"}},{"@type":"WebPage","@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#webpage","url":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/","name":"\u041a\u0435\u0439\u0441: Qwen3.8-Flash-Next \u2014 MoE 125B\/6B \u043d\u0430 DGX Spark","description":"26 \u0430\u0432\u0433\u0443\u0441\u0442\u0430 2026 \u0433\u043e\u0434\u0430 Alibaba \u0432\u044b\u043f\u0443\u0441\u0442\u0438\u043b\u0430 Qwen3.8-Flash-Next \u2014 \u044d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442\u0430\u043b\u044c\u043d\u0443\u044e \u043c\u043e\u0434\u0435\u043b\u044c \u0441 \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u043e\u0439, \u043a\u043e\u0442\u043e\u0440\u0443\u044e \u043a\u043e\u043c\u0430\u043d\u0434\u0430 \u043d\u0430\u0437\u044b\u0432\u0430\u0435\u0442 \u00ab\u043f\u0440\u0435\u0432\u044c\u044e Qwen4\u00bb. 125 \u043c\u043b\u0440\u0434 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432, \u043d\u043e \u043d\u0430 \u043a\u0430\u0436\u0434\u043e\u043c \u0448\u0430\u0433\u0435 \u0430\u043a\u0442\u0438\u0432\u043d\u044b \u0442\u043e\u043b\u044c\u043a\u043e 6 \u043c\u043b","inLanguage":"ru-RU","isPartOf":{"@id":"https:\/\/ip-calculator.ru\/blog\/#website"},"breadcrumb":{"@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#breadcrumblist"},"author":{"@id":"https:\/\/ip-calculator.ru\/blog\/author\/anatolyi\/#author"},"creator":{"@id":"https:\/\/ip-calculator.ru\/blog\/author\/anatolyi\/#author"},"image":{"@type":"ImageObject","url":"https:\/\/ip-calculator.ru\/blog\/wp-content\/uploads\/2026\/08\/kejs-qwen38-flash-next-moe-617.jpg","@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#mainImage","width":2400,"height":1800},"primaryImageOfPage":{"@id":"https:\/\/ip-calculator.ru\/blog\/artificial-intelligence\/kejs-qwen38-flash-next-moe\/#mainImage"},"datePublished":"2026-09-05T13:00:00+03:00","dateModified":"2026-08-30T03:35:53+03:00"},{"@type":"WebSite","@id":"https:\/\/ip-calculator.ru\/blog\/#website","url":"https:\/\/ip-calculator.ru\/blog\/","name":"\u0421\u0435\u0442\u044c \u0431\u0435\u0437 \u043f\u0440\u043e\u0431\u043b\u0435\u043c","description":"\u041d\u0430\u0441\u0442\u0440\u043e\u0439\u043a\u0430 \u0438 \u043e\u0431\u0441\u043b\u0443\u0436\u0438\u0432\u0430\u043d\u0438\u0435 \u0441\u0435\u0442\u0438","inLanguage":"ru-RU","publisher":{"@id":"https:\/\/ip-calculator.ru\/blog\/#person"}}]}},"aioseo_meta_data":{"post_id":"47841","title":"\u041a\u0435\u0439\u0441: Qwen3.8-Flash-Next \u2014 MoE 125B\/6B \u043d\u0430 DGX Spark","description":"26 \u0430\u0432\u0433\u0443\u0441\u0442\u0430 2026 \u0433\u043e\u0434\u0430 Alibaba \u0432\u044b\u043f\u0443\u0441\u0442\u0438\u043b\u0430 Qwen3.8-Flash-Next \u2014 \u044d\u043a\u0441\u043f\u0435\u0440\u0438\u043c\u0435\u043d\u0442\u0430\u043b\u044c\u043d\u0443\u044e \u043c\u043e\u0434\u0435\u043b\u044c \u0441 \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u043e\u0439, \u043a\u043e\u0442\u043e\u0440\u0443\u044e \u043a\u043e\u043c\u0430\u043d\u0434\u0430 \u043d\u0430\u0437\u044b\u0432\u0430\u0435\u0442 \u00ab\u043f\u0440\u0435\u0432\u044c\u044e Qwen4\u00bb. 125 \u043c\u043b\u0440\u0434 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432, \u043d\u043e \u043d\u0430 \u043a\u0430\u0436\u0434\u043e\u043c \u0448\u0430\u0433\u0435 \u0430\u043a\u0442\u0438\u0432\u043d\u044b \u0442\u043e\u043b\u044c\u043a\u043e 6 \u043c\u043b","keywords":null,"keyphrases":null,"primary_term":null,"canonical_url":null,"og_title":null,"og_description":null,"og_object_type":"default","og_image_type":"default","og_image_url":null,"og_image_width":null,"og_image_height":null,"og_image_custom_url":null,"og_image_custom_fields":null,"og_video":"","og_custom_url":null,"og_article_section":null,"og_article_tags":null,"twitter_use_og":false,"twitter_card":"default","twitter_image_type":"default","twitter_image_url":null,"twitter_image_custom_url":null,"twitter_image_custom_fields":null,"twitter_title":null,"twitter_description":null,"schema":{"blockGraphs":[],"customGraphs":[],"default":{"data":{"Article":[],"Course":[],"Dataset":[],"FAQPage":[],"Movie":[],"Person":[],"Product":[],"ProductReview":[],"Car":[],"Recipe":[],"Service":[],"SoftwareApplication":[],"WebPage":[]},"graphName":"Article","isEnabled":true},"graphs":[]},"schema_type":"default","schema_type_options":null,"pillar_content":false,"robots_default":true,"robots_noindex":false,"robots_noarchive":false,"robots_nosnippet":false,"robots_nofollow":false,"robots_noimageindex":false,"robots_noodp":false,"robots_notranslate":false,"robots_max_snippet":"-1","robots_max_videopreview":"-1","robots_max_imagepreview":"large","priority":null,"frequency":"default","location":null,"local_seo":null,"breadcrumb_settings":null,"limit_modified_date":false,"ai":null,"created":"2026-08-30 00:07:08","updated":"2026-08-30 00:07:14","focus_keyword":null,"additional_keywords":null,"truseo_locale":null,"seo_analyzer_scan_date":null},"_links":{"self":[{"href":"https:\/\/ip-calculator.ru\/blog\/wp-json\/wp\/v2\/posts\/47841","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ip-calculator.ru\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ip-calculator.ru\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ip-calculator.ru\/blog\/wp-json\/wp\/v2\/users\/5"}],"replies":[{"embeddable":true,"href":"https:\/\/ip-calculator.ru\/blog\/wp-json\/wp\/v2\/comments?post=47841"}],"version-history":[{"count":1,"href":"https:\/\/ip-calculator.ru\/blog\/wp-json\/wp\/v2\/posts\/47841\/revisions"}],"predecessor-version":[{"id":47842,"href":"https:\/\/ip-calculator.ru\/blog\/wp-json\/wp\/v2\/posts\/47841\/revisions\/47842"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/ip-calculator.ru\/blog\/wp-json\/wp\/v2\/media\/47843"}],"wp:attachment":[{"href":"https:\/\/ip-calculator.ru\/blog\/wp-json\/wp\/v2\/media?parent=47841"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ip-calculator.ru\/blog\/wp-json\/wp\/v2\/categories?post=47841"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ip-calculator.ru\/blog\/wp-json\/wp\/v2\/tags?post=47841"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}