Are the weights open and can I run this locally?

Yes, weights are published on Hugging Face at Qwen/Qwen3.8-2.4T-A95B and are compatible with vLLM and SGLang, but 2.4 trillion total parameters means you need substantial multi-GPU infrastructure to self-host at full precision.

What does it cost and how long is the context?

Via API the price is $2.00 per million input tokens and $6.00 per million output tokens; native context is 262,144 tokens, extensible to around one million.

How does this differ from Qwen3.8 Max?

Qwen3.8 Max is the managed API version built on the same model and adds vision input, non-thinking mode, a 1M context by default, and official built-in tools; those features are not part of this open-weight release.

Qwen3.8 2.4T A95B

1539°

#llm #qwen #reasoning #coding #agents #research

Run this model

Offers / Ads

OpenRouter Terminal / API

Pay per token

Chat with this model in the browser or call it from your code with one API key.

Qwen | 1,048,576 token context | $2.00/M input, $6.00/M output. Qwen3.8 2.4T A95B is an open-weight sparse mixture-of-experts model from Qwen and the open-weight variant of Qwen3.8 Max, with 95 billion active parameters out of 2.4 trillion total. It is suited for coding, research, complex reasoning, and agentic workflows.

Guides for this model

Ollama vs LM Studio (2026): which local AI app to use, and which top models each one runs

Best local LLM by use case (October 2026): the right open model for every job

How much VRAM do I actually need for local AI?

What is a GGUF file? Quantized models explained (Q4, Q8, fp8, NF4)

All guides

Captain's notes

Written by Captain, the house editor

Notes from 2026-10-13

Qwen3.8 2.4T A95B is Qwen's largest open-weight release to date: a sparse mixture-of-experts model with 2.4 trillion total parameters and 95 billion active at inference. It is the open-weight counterpart to the closed Qwen3.8 Max, sharing the same architecture but without the managed API's vision input and built-in tools. If you need frontier-class reasoning, coding, or multi-step agentic work and want weights you can run yourself, this is the one to look at.

Known for

2.4T total parameters with only 95B active per forward pass, keeping inference costs lower than dense equivalents
Long-horizon agentic task completion with stronger autonomous planning and environment-feedback handling
Tunable reasoning depth via reasoning_effort, with thinking context preserved across turns
Native 262K context, extensible to roughly 1 million tokens
Open weights compatible with vLLM, SGLang and similar serving frameworks

Watch out for

2.4 trillion total parameters means serious hardware to self-host; most users will hit an API instead
The managed Qwen3.8 Max adds vision input and built-in tools that this open release does not include
At $6.00 per million output tokens via API, costs climb fast on long agentic runs

Best for

Prompts to try

Debug a concurrency bug

Here is a Python async function that occasionally raises a RuntimeError about an event loop being closed. Walk through what is likely causing it, then rewrite the function so it handles the event loop lifecycle correctly. Show your reasoning before the final code. ```python async def fetch_all(urls): loop = asyncio.get_event_loop() tasks = [loop.create_task(fetch(u)) for u in urls] return await asyncio.gather(*tasks) ```

Plan a multi-step research task

You are working autonomously. Your goal: produce a structured literature summary on the current state of sparse mixture-of-experts scaling laws. Break the task into concrete steps, identify what information you would retrieve at each step, flag any ambiguities that would require clarification before proceeding, and outline what the final output should contain.

Reason through a logic puzzle

Five engineers join a project on different days Monday through Friday. Alice joins before Bob. Carol joins the day after Dave. Eve joins on Wednesday. Bob does not join on Friday. Dave joins on Monday. What day does each person join? Show your deduction step by step before giving the final answer.

Compare with

Qwen3.7 Flash
API-only model
1,000,000 token context · $0.03/M input, $0.13/M output
Qwen3.8 Max (0902)
API-only model
1,000,000 token context · $2.00/M input, $6.00/M output
Qwen3.8 Omni Flash
API-only model
1,000,000 token context · $0.15/M input, $0.47/M output
Qwen3 30B A3B Thinking 2507
API-only model
81,920 token context · $0.20/M input, $2.40/M output

More text models from qwen

Questions people ask

Are the weights open and can I run this locally?: Yes, weights are published on Hugging Face at Qwen/Qwen3.8-2.4T-A95B and are compatible with vLLM and SGLang, but 2.4 trillion total parameters means you need substantial multi-GPU infrastructure to self-host at full precision.
What does it cost and how long is the context?: Via API the price is $2.00 per million input tokens and $6.00 per million output tokens; native context is 262,144 tokens, extensible to around one million.
How does this differ from Qwen3.8 Max?: Qwen3.8 Max is the managed API version built on the same model and adds vision input, non-thinking mode, a 1M context by default, and official built-in tools; those features are not part of this open-weight release.

Captain's verdict
The most capable open-weight model Qwen has shipped, worth serious consideration for coding agents and complex reasoning if you can absorb the infrastructure or API cost.

Model card on Hugging Face

Captain writes these notes from the model's own description and model card. Prices and context sizes are the provider's at the time of writing and can change.

Ranking badge

Updated Oct 22, 2026

Made this model? Show this badge on your page.

Paste it into your Hugging Face card or Civitai description.

Change colors / personalize / get code

Not this creator?

Get your own badge

How this score adds up

This model has no popularity data from any source yet, so it carries a stand-in number below 7,000° that keeps the lists complete. It never ranks above a model with data. The breakdown appears here as soon as a source reports it.

Glossary: the terms used here

Heat score: One number for how much a model is used and wanted right now, from 0° to 9,999°. Every list on this site that says hottest is sorted by it and nothing else.
Floor: The 7,000° every model with real data starts from. The scale runs from 0° to 9,999°; the part below the floor is reserved for stand-ins (models without data), so that a model with data always ranks above one without. The floor is the same for everyone and decides nothing; the degrees above it are the measurement.
Download-equivalents: Everything the score treats as a download, added up: downloads themselves, each torrent seeder as 1,000, each AI Horde image and each run on a generation site as one.
Log scale: Counting by tens, not by ones: going from 1,000 to 10,000 downloads adds the same degrees as going from 10,000 to 100,000. It keeps giants from crowding out everything else.
Likes: Thumbs up and likes from every source, worth half a download each on the log scale.
Adoption rate: How many downloads a day a model gets right now, from the difference between two readings of a source. A quarter of the weight of the download term.
Measured and estimated: Measured means two readings of a source exist and the difference is real. Estimated means only today's total is known and the past is spread over the model's life; the site says which one it is showing.
Seeder: Someone who keeps a copy of the weights online for others to download over a torrent, or a hosted mirror doing the same. Each one counts as 1,000 downloads.
AI Horde: A volunteer network that generates images for free on donated computers. The number of images it has made with a model is public, and each counts as one download.
Generation sites: Sites that host their own copies of a model and let people generate with it, such as PixAI, SeaArt and Yodayo. Their downloads count like Civitai's and each generation run as one download.
OpenRouter tokens: A text model served through an API, such as Claude or GPT, has no weights to download, so its usage is measured in tokens: the pieces of text it read and wrote through OpenRouter, a gateway many apps use, over the last 30 days. Every 10,000,000 tokens count as one download. Source: OpenRouter (openrouter.ai/rankings), CC BY 4.0.
Full model (checkpoint): The weights themselves, the file that generates on its own. It gets a bonus because add-ons depend on it.
LoRA and embedding: Small add-on files that steer someone else's weights toward a style, a character or a concept. They cannot run alone.
Family and weights: The base a model is built on, such as SD 1.5, SDXL, Flux or Wan. Every model inherits the release date of its family's weights.
Age: Degrees taken off for how long ago the family's weights came out: 150° a year, up to 750°. It keeps old downloads that had years to pile up from freezing the top of the list.
Family boost: A small, bounded nudge for LoRAs of a family this site is pushing forward right now, always listed in the breakdown.
Stand-in score: A stable number below the floor given to a model no source reports on yet, so lists stay complete. It never ranks above a model with data.
New! (grace period): A model less than 7 days old shows New! instead of a score and holds no place in the rankings. Its score is computed from day one, but a model one day old grows faster than anything, so without the wait the week's #1 would always be whatever came out yesterday rather than what held up for a week.
Trend: How a model's growth over a period compares with the typical model's growth over the same period. Faster than typical reads as a rise, slower as a fall.
Typical model: The middle of the catalog: half the models with a figure grew faster over the period, half slower.

Similar Models

AI Models for Image, Video & Text Generation

Qwen3.8 2.4T A95B is a text model for the Text / LLM family listed on ModelTrends.ai, a read-only catalog of open source AI image, video and text models. Compare it with other Text / LLM models and check its heat score to see how it is trending.

More Text / LLM models llm qwen reasoning coding agents research long context

Browse by Model Family