# Agnes-AI/Agnes-3.0-Qwen (QFN finetune) — retained record

> Retained reddit record from reddit:LocalLLaMA.

Canonical: https://highsignal.app/data/records/c7ded4dc75ee3a6b

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:LocalLLaMA)  published 2026-10-20 17:11 ingested 2026-10-21 02:41

[original source ↗](https://www.reddit.com/r/LocalLLaMA/comments/1xaxz80/agnesaiagnes30qwen_qfn_finetune/)

## retained text



```
  
[https://huggingface.co/bartowski/Agnes-AI\_Agnes-3.0-Qwen-GGUF](https://huggingface.co/bartowski/Agnes-AI_Agnes-3.0-Qwen-GGUF)

* **Strong agentic performance:** agnes-3.0-qwen scores highest in its comparison set on τ³-Banking, and is within about one point of the best model on Terminal-Bench 2.1, while being a much smaller model than most of the flagship systems it is compared against.
* **Reliable instruction following:** on IFBench, agnes-3.0-qwen outperforms GPT-5.6 Sol (Max) and Claude Opus 4.8 (Max).
* **Built for demanding work:** a **1M-token context window**, extended reasoning with selectable **reasoning effort** (`xhigh`, `medium`, `low`), tool calling, and text, image, and video understanding.

# Model Information

|Property|Value|
|:-|:-|
|Developed by|Agnes AI|
|Model name|agnes-3.0-qwen|
|Model type|Multimodal reasoning model (hybrid attention MoE)|
|Total parameters|\~180B (512 experts, 10 active per token)|
|License|[Qwen Community License 1.0](https://huggingface.co/Agnes-AI/Agnes-3.0-Qwen/tree/main/LICENSE)|
|Languages|English, Chinese|
|Context window|1,048,576 tokens|
|Maximum output|65,536 tokens|
|Input modalities|Text, image, video|
|Output modality|Text|
|Precision|BF16|
|Reasoning effort|`xhigh` (default), `medium`, `low`|
|Tool calling|Yes|
|Streaming|Yes|
|Release date|October 2026|
```



document fetched 2026-10-21 00:41 · canonical https://reddit.com/r/LocalLLaMA/comments/1xaxz80/agnesaiagnes30qwen_qfn_finetune
