# Introducing DeepSeek-V4.1-Flash-1017 Dynamic v3.1 Unsloth GGUFs — retained record

> Retained reddit record from reddit:LocalLLaMA.

Canonical: https://highsignal.app/data/records/49befce01370d824

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:LocalLLaMA)  published 2026-10-30 13:48 ingested 2026-10-31 02:40

[original source ↗](https://www.reddit.com/r/LocalLLaMA/comments/1xjfuqh/introducing_deepseekv41flash1017_dynamic_v31/)

## retained text



```
Hey everyone! We’re releasing new DeepSeek-V4.1-1017 Dynamic V3.1 (preview) GGUFs which showcases how Unsloth’s GGUFs retain the **most accuracy** for its size across **7 different benchmarks**.  
  
We benchmarked ISTA's GSQ-RCO GGUFs but could **not reproduce their reported results**, even when using the highest scores. The biggest discrepancy was on AIME25 where they claimed 100% accuracy recovery to baseline, but was only \~91.88% in benchmark tests.

Across **AIME25**, **LiveCodeBench**, **GPQA Diamond**, **Aider Polyglot**, and more, Unsloth Dynamic V3.1 **outperforms** GSQ-RCO GGUFs on nearly all metrics. Even our old v3 does better on nearly all metrics than ISTA.

We ran each benchmark at least 3 times (AIME25 was 16 times) and averaged the results to ensure a fair, reliable comparison. More results coming next week! All benchmark data, methodologies and results are in our blog.

Separately, for KLD, there were queries whether it’s a reliable measure of accuracy. The primary goal of quantization is to minimize KLD but also minimize disk space. If you get a KLD of 0, this means your quantized model == 16bit model, as logits are perfectly modelled. Our benchmarks show on KLD and top-1% across multiple domains (coding, agentic coding, chat etc), UDv3.1 does much better than GSQ-RCO across every domain.

We show that lower bit quants do think **somewhat longer than the unquantized model** \- we plan to investigate and make quants even better. However, we're still much ahead vs GSQ-RCO on even reasoning length as well, and provide plots for truncated answers in our blog.

The model will be able to run on 128GB RAM/unified memory devices or even on 64GB RAM with engram offloading.

Blog with all details and more benchmarks:[ ](https://unsloth.ai/docs/basics/dynamic-3.0-ggufs)[https://unsloth.ai/docs/models/deepseek-v4.1-flash](https://unsloth.ai/docs/models/deepseek-v4.1-flash)

GGUF:[ ](https://huggingface.co/unsloth/Qwen3.8-27B-GGUF)[https://huggingface.co/unsloth/DeepSeek-V4.1-Flash-1017-GGUF](https://huggingface.co/unsloth/DeepSeek-V4.1-Flash-1017-GGUF)

We also introduced lots of new things to Unsloth Desktop including Browser Use, auto compaction, a whole new TTS pipeline (with voice cloning, separation etc), sandboxing, and background customization and more:[ https://github.com/unslothai/unsloth](https://github.com/unslothai/unsloth)

Hope you all have a Happy Halloween!

Also table of numbers below which is also described with more plots and data in our blog:

|Quant|Size GB|AIME25|HMMT Feb26|GPQA-D|MMLU-Pro|IFEval|LCB v6 (32K)|Avg of 6|Aider pass@1|
|:-|:-|:-|:-|:-|:-|:-|:-|:-|:-|
|UD-Q8\_K\_XL|508.4|97.08|68.94|90.47|88.61|95.10|85.90|87.68|\-|
|UD-Q2\_K\_XL|334.8|97.50|69.89|89.58|88.28|94.92|84.76|87.49|\-|
|UD-IQ2\_XXS|300.5|96.88|67.61|88.95|87.53|95.19|83.62|86.63|44.38|
|UD-IQ1\_M|250.4|95.00|62.88|88.38|87.53|95.10|84.38|85.55|42.13|
|UD-IQ1\_S|233.0|92.92|57.01|87.63|86.62|94.92|82.10|83.53|36.52|
|GSQ-RCO IQ2\_XS|265.8|95.00|64.39|88.64|86.62|94.27|84.95|85.64|41.57|
|GSQ-RCO IQ2\_XXS|248.8|95.21|57.95|85.80|86.53|93.72|82.86|83.68|36.52|
|GSQ-RCO IQ1\_M|231.3|91.88|53.60|85.10|85.49|91.77|82.86|81.78|29.78|
```



document fetched 2026-10-31 00:45 · canonical https://reddit.com/r/LocalLLaMA/comments/1xjfuqh/introducing_deepseekv41flash1017_dynamic_v31
