Trinity Mini
- Context
- 262Ktokens
- Released
- 1y ago
- Size
- 80Bparameters
Qwen3-Next 80B-A3B (Thinking) is an open-weights model from Alibaba and the inaugural model of the Qwen3-Next architecture, a line the company introduces as a new generation of ultra-efficient model design. The launch announcement centers on computational efficiency through a hybrid attention mechanism and a highly sparse Mixture-of-Experts design, with Alibaba claiming more than 10x higher inference throughput than its dense Qwen3-32B model on long-context inputs. The Thinking variant is built for complex reasoning tasks.
- Language model
- Open weights · 80B
- Reasoning
- Tools
How it compares
Size class
Medium80BAn open-weight model of 25 to 100 billion parameters, for a workstation or a single server GPU.
Open models of a similar size from other labs
Vs. your stack
Add the models you use to see how this one compares.
Sign inAdditional source evidence
- Source model
- alibaba/qwen3-next-80b-a3b-thinking
- Source family
- qwen
- Last updated
- 2025-09
- Knowledge cutoff
- 2025-04
- Model context
- 262K
- Model output limit
- 33K
- Open weights
- Yes
- File attachments
- No
- Reasoning
- Yes
- Tool calling
- Yes
- Temperature control
- Yes
Downloadable weights
Complete source record (JSON)
{
"id": "alibaba/qwen3-next-80b-a3b-thinking",
"name": "Qwen3-Next 80B-A3B (Thinking)",
"description": "Efficient Qwen thinking model for local reasoning, math, and coding agents",
"family": "qwen",
"last_updated": "2025-09",
"knowledge": "2025-04",
"attachment": false,
"reasoning": true,
"tool_call": true,
"temperature": true,
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"limit": {
"context": 262144,
"output": 32768
},
"release_date": "2025-09",
"open_weights": true,
"weights": [
{
"label": "Hugging Face",
"url": "https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Thinking"
}
]
}Frequently asked questions
Running locally
Qwen3-Next 80B-A3B (Thinking) needs an estimated 56.3 GB VRAM at Q4, 98.2 GB at Q8 and 184.5 GB at FP16. These estimates are based on the 80B parameter count and cover the weights plus about 15% and 0.5 GB for a short context.
No. Qwen3-Next 80B-A3B (Thinking) does not fit on a 24 GB, 16 GB, 12 GB or 8 GB GPU at Q4, Q8 or FP16 (estimate).
Yes, at Q4. Q4 needs an estimated 56.3 GB, and assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits Q4 is 96 GB (estimate).
Privacy and data
Yes. Qwen3-Next 80B-A3B (Thinking) has open weights, so you can self-host it for full control of your data.
Capabilities and specs
The context window of Qwen3-Next 80B-A3B (Thinking) is 262K tokens (262,144), and its maximum output is 33K tokens (32,768).
No. The only supported input modality of Qwen3-Next 80B-A3B (Thinking) is text.
Yes. Qwen3-Next 80B-A3B (Thinking) supports tool calling (function calling).
Yes. Qwen3-Next 80B-A3B (Thinking) supports a reasoning mode.
The knowledge cutoff of Qwen3-Next 80B-A3B (Thinking) is 2025-04.
Quality and benchmarks
Qwen3-Next 80B-A3B (Thinking) ranks 203 on Arena Text (Arena's own rank), with a score of 1369.
The closest alternatives, matched on model kind and release date (no comparable price), are ERNIE 4.5 21B A3B Thinking by Baidu, Command A Reasoning by Cohere and MolmoAct 7B-D by Ai2.
ERNIE 4.5 21B A3B ThinkingQwen3-Next 80B-A3B (Thinking) vs ERNIE 4.5 21B A3B ThinkingCommand A ReasoningQwen3-Next 80B-A3B (Thinking) vs Command A ReasoningMolmoAct 7B-DQwen3-Next 80B-A3B (Thinking) vs MolmoAct 7B-D
Versions and lineage
Qwen3-Next 80B-A3B (Thinking) has a 262K token (262,144) context and was released September 1, 2025, while Qwen3 235B-A22B has a 131K token (131,072) context and was released April 28, 2025.
Qwen3 235B-A22BQwen3-Next 80B-A3B (Thinking) vs Qwen3 235B-A22B
Licensing and access
Qwen3-Next 80B-A3B (Thinking) is an open weights model.
Basics
Qwen made Qwen3-Next 80B-A3B (Thinking), which was released on September 1, 2025.
Yes. Qwen3-Next 80B-A3B (Thinking) has open weights.
Generated from Vertical Tabs data, updated November 11, 2026.
Discussion
No comments yet
Be the first to share your thoughts.