--- title: Qwen3.8-2.4T | daily.dev description: Qwen3.8-2.4T-A95B-FP8 is an FP8-quantized release of Qwen's largest open model to date, with 2.4T total parameters and 95B activated via a mixture-of-experts... canonical: https://daily.dev/posts/qwen3-8-2-4t-c7kqr4bkk twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Qwen3.8-2.4T | daily.dev og:description: Qwen3.8-2.4T-A95B-FP8 is an FP8-quantized release of Qwen's largest open model to date, with 2.4T total parameters and 95B activated via a mixture-of-experts... og:url: https://daily.dev/posts/qwen3-8-2-4t-c7kqr4bkk og:image: https://api.daily.dev/og/posts/C7kqr4bKk.png og:image:alt: Qwen3.8-2.4T og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Qwen3.8-2.4T **[Hacker News](https://daily.dev/sources/hn)** · 11 min read · 0 upvotes · 0 comments ## Summary Qwen3.8-2.4T-A95B-FP8 is an FP8-quantized release of Qwen's largest open model to date, with 2.4T total parameters and 95B activated via a mixture-of-experts architecture. Built on the Qwen3.5 architecture, it targets coding, agentic, and long-horizon tasks, supporting up to 1M token context, tunable reasoning effort, and thinking-mode-only inference. Benchmark tables compare it against Claude Opus 4.8, Claude Fable 5, and GPT-5.6 Sol across coding agent, general agent, and general capability suites, with usage instructions for Transformers, vLLM, SGLang, and Docker. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What are the specs of the Qwen3.8-2.4T-A95B model architecture? Qwen3.8-2.4T-A95B is a mixture-of-experts causal language model with 2.4T total parameters and 95B activated per token. It has 92 layers, a hidden dimension of 8192, 512 experts with 10 routed plus 1 shared activated per token, and combines Gated DeltaNet with Gated Attention blocks. Native context length is 262,144 tokens, extensible to 1,010,000 tokens. _daily.dev helps engineers sizing and deploying massive MoE models compare architectures like this one._ ### How do I run Qwen3.8-2.4T-A95B-FP8 with vLLM? Install vLLM via pip install vllm, then start the server with vllm serve "Qwen/Qwen3.8-2.4T-A95B-FP8". Once running, call it through the OpenAI-compatible chat completions endpoint at http://localhost:8000/v1/chat/completions with a standard JSON payload containing the model name and messages array. _Track new model serving workflows like this one for vLLM on daily.dev as you deploy large open models._ ### Can I disable thinking mode when using the Qwen3.8-2.4T-A95B model? No, thinking mode cannot be disabled for Qwen3.8-2.4T-A95B; it is a text-only model that requires reasoning for every interaction. Every response automatically starts with reasoning wrapped in ... tags before the final output, and reasoning depth can instead be tuned via the reasoning_effort parameter (xhigh, medium, or low). _daily.dev keeps developers current on model configuration quirks like mandatory thinking modes._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 18 comments across hackernews (as of 2026-08-13). **TL;DR:** Discussion focuses mostly on confusion over naming and identity (is this the same as a smaller variant?) and skepticism about who could actually run a model this massive, rather than praise or criticism of its capabilities. **Sentiment:** 10% positive · 30% mixed · 60% skeptical **The pushback** - Doubt that anyone outside major labs or huge corporations has the hardware to run a model this size. - Confusion and suspicion around naming, with some claiming it's actually a differently-branded release. **By community** - hackernews (skeptical): Commenters question the model's true identity/naming and doubt practical feasibility of running something this large outside big labs. **Hottest debate:** Whether this release is genuinely a new, larger model or just a re-branded/renamed version of a smaller one. **Open questions** - Is this truly a distinct model from the smaller '3.8-27B' variant or the same lineage rebranded? - Who realistically has the hardware to run a model of this size? **Highlights** > I wonder besides big labs and REALLY big corporations who can run the 2.6TB q8... I mean, the 1 bit one is 508GB. Assuming 256k context size as irrelevant at those sizes, I would need 22 AMD 7900XTX (24GB vram) to run this for the 1 bit one, and 113 for the 2.6TB 113 and assumming close to constant load and the gpus taking turns as it does on my machine with two gpus is around 113gpus*113W = 11KW... just to hold and run that one instance... and only god knows the cooling requirements... I guess just someone on Meta/Googly/Claudy will say... oh nice, let's download it and run it... How did the unsloth guy did to process a file that size? https://huggingface.co/unsloth/Qwen3.8-2.4T-A95B-GGUF > — [xlayn on hackernews · 1 comments](https://news.ycombinator.com/item?id=49276897) > this is Qwen3.8 max, right? https://qwen.ai/blog?id=qwen3.8 > — [brrrrrm on hackernews · 2 comments](https://news.ycombinator.com/item?id=49276436) > they explicitly say this is qwen3.8 max in the hugging face page > — [efficax on hackernews](https://news.ycombinator.com/item?id=49276829) > We're all really waiting for the 3.8-27B which is due out in 2 days. > — [UncleOxidant on hackernews · 1 comments](https://news.ycombinator.com/item?id=49276464) **Source threads** - [hackernews](https://news.ycombinator.com/item?id=49274950) · 79 points · 18 comments ## Similar posts on daily.dev - [Qwen3.6–35B-A3B: The Most Practical Open-Source AI Model Yet?](https://daily.dev/posts/qwen3-6-35b-a3b-the-most-practical-open-source-ai-model-yet--eihfhk9n5) · Faun · 77 upvotes · 6 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning), [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#vllm](https://daily.dev/tags/vllm), [#qwen](https://daily.dev/tags/qwen) [View this post on daily.dev](https://daily.dev/posts/qwen3-8-2-4t-c7kqr4bkk) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Qwen3.8-2.4T","url":"https://daily.dev/posts/qwen3-8-2-4t-c7kqr4bkk","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/qwen3-8-2-4t-c7kqr4bkk"},"datePublished":"2026-08-12T18:57:17.858Z","dateModified":"2026-08-13T03:03:18.886Z","description":"Qwen3.8-2.4T-A95B-FP8 is an FP8-quantized release of Qwen's largest open model to date, with 2.4T total parameters and 95B activated via a mixture-of-experts...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/fc6a97c449539bbe8d75543fb9c79780?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/fc6a97c449539bbe8d75543fb9c79780?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Hacker News","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Hacker News","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/hn","url":"https://daily.dev/sources/hn"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/qwen3-8-2-4t-c7kqr4bkk","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"machine-learning,llm,ai-agents,vllm,qwen","timeRequired":"PT11M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Hacker News","item":"https://daily.dev/sources/hn"},{"@type":"ListItem","position":3,"name":"Qwen3.8-2.4T"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/qwen3-8-2-4t-c7kqr4bkk#faq","mainEntity":[{"@type":"Question","name":"What are the specs of the Qwen3.8-2.4T-A95B model architecture?","acceptedAnswer":{"@type":"Answer","text":"Qwen3.8-2.4T-A95B is a mixture-of-experts causal language model with 2.4T total parameters and 95B activated per token. It has 92 layers, a hidden dimension of 8192, 512 experts with 10 routed plus 1 shared activated per token, and combines Gated DeltaNet with Gated Attention blocks. Native context length is 262,144 tokens, extensible to 1,010,000 tokens. daily.dev helps engineers sizing and deploying massive MoE models compare architectures like this one."}},{"@type":"Question","name":"How do I run Qwen3.8-2.4T-A95B-FP8 with vLLM?","acceptedAnswer":{"@type":"Answer","text":"Install vLLM via pip install vllm, then start the server with vllm serve \"Qwen/Qwen3.8-2.4T-A95B-FP8\". Once running, call it through the OpenAI-compatible chat completions endpoint at http://localhost:8000/v1/chat/completions with a standard JSON payload containing the model name and messages array. Track new model serving workflows like this one for vLLM on daily.dev as you deploy large open models."}},{"@type":"Question","name":"Can I disable thinking mode when using the Qwen3.8-2.4T-A95B model?","acceptedAnswer":{"@type":"Answer","text":"No, thinking mode cannot be disabled for Qwen3.8-2.4T-A95B; it is a text-only model that requires reasoning for every interaction. Every response automatically starts with reasoning wrapped in ... tags before the final output, and reasoning depth can instead be tuned via the reasoning_effort parameter (xhigh, medium, or low). daily.dev keeps developers current on model configuration quirks like mandatory thinking modes."}}]} ```