Unsloth GGUF quants
Hacker ApprovedQuantized model in AI tools
Widely recommended GGUF quantizations of popular open models, valued for quick releases and low memory use, though some early releases had bugs and errors.
33 people recommended Unsloth GGUF quants in Hacker News comments, 2 had mixed feelings and 10 criticized it, November 2025 to October 2026.
Summary written by Claude from 48 comments.
45 people gave an opinion
- recommend it
- 33
- mixed feelings
- 2
- warn against or regret it
- 10
The record
Recommended in 10 of 12 months
In 28 different threads.
- November 2025: recommended
- December 2025: not recommended
- January 2026: not recommended
- February 2026: recommended
- March 2026: recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: recommended
Why it carries the badge
- 33 people recommend itNeeds 10 or more, met.
- 3.3 recommenders for every criticNeeds 2 to 1 or better, met.
- Recommended in 10 of 12 monthsNeeds 4 or more, met.
- Recommended in 28 threadsNeeds 5 or more, met.
- The biggest thread supplies 12% of its recommendersNeeds half or less, met.
The same rules apply to every product, and the badge can’t be bought. Mixed feelings count toward neither side. How the badge works
What people say about it
Reasons grouped from a sample of 40 of the 49 opinions about it, picked in proportion to how many people recommend it, have mixed feelings and criticize it. Each mark is one comment in the sample. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Reliable, go-to quants
“The unsloth quants seem to have been fixed, so they are probably the go-to again”
Comment from 28 Feb 2026 in “Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers” “I have had great luck with unsloth quants so far.”
Comment from 13 Aug 2026 in “Qwen/Qwen3.8-2.4T-A95B” “They are the most reliable in my experience”
Comment from 12 Aug 2026 in “Qwen/Qwen3.8-2.4T-A95B” - Comment from 3 Apr 2026 in “Qwen3.6-Plus: Towards Real World Agents”
- Comment from 17 Sep 2026 in “Breaking the 1.58-bit Barrier for Ternary LLMs”
- Comment from 28 Feb 2026 in “How do I cancel my ChatGPT subscription?”
- Comment from 15 Aug 2026 in “Qwen 3.8 27B is out: open weights, best local dense model yet”
- Comment from 2 Oct 2026 in “From the creator of Redis; run LLM locally with ds4”
Runs on low-memory hardware
“something like Unsloth's UD q8 XL or q6 XL quants free up a lot of memory and bandwidth”
Comment from 19 Apr 2026 in “My first impressions on the ROCm and Strix Halo” “amazed by the Unsloth team by how much they can reduce the requirements”
Comment from 29 Oct 2026 in “We're releasing the weights of Mistral Large 4” “the unsloth versions can run on a potato and are actually useful”
Comment from 20 May 2026 in “Qwen3.7-Max: The Agent Frontier” - Comment from 14 Aug 2026 in “Qwen3.8-27B”
- Comment from 18 Sep 2026 in “Bonsai 2 27B: Near-Lossless Compression in a 9x Smaller Footprint”
Better quality than self-made quants
“They are higher quality than the quants you can make yourself due to their imatrix datasets”
Comment from 19 Apr 2026 in “My first impressions on the ROCm and Strix Halo” “Unsloth's quants are better than the original Google's QAT as posted in the article.”
Comment from 5 Jun 2026 in “Gemma 4 QAT models: Optimizing compression for mobile and laptop efficiency” “we're now the best in terms of KLD and disk space”
Comment from 16 Apr 2026 in “Qwen3.6-35B-A3B: Agentic Coding Power, Now Open to All” Tested, quickly published quants
“they make it first, then they run a lot of tests, KD, perplexity, etc, they publish it”
Comment from 12 Aug 2026 in “Qwen/Qwen3.8-2.4T-A95B” “Unsloth are doing a fantastic job of publishing quants of the major models quickly”
Comment from 14 May 2026 in “What's in a GGUF, besides the weights – and what's still missing?” “unsloth produces nice quants for it and even provides guidelines”
Comment from 27 Mar 2026 in “$500 GPU outperforms Claude Sonnet on coding benchmarks”
What people warn about
Each mark is one comment
Errors and bugs in releases
“I had several issues with unsloth gguf, even for models released a few months back like gemma 4”
Comment from 12 Aug 2026 in “Qwen/Qwen3.8-2.4T-A95B” “Sometimes unsloth has broken ones for a particular model, sometimes no quants at all”
Comment from 6 Mar 2026 in “Something is afoot in the land of Qwen” “Running unsloth versions with llama.cpp, constant issues.”
Comment from 30 Jun 2026 in “Qwen 3.6 27B is the sweet spot for local development” - Comment from 26 Aug 2026 in “Qwen3.8-Flash-Next”
- Comment from 28 Feb 2026 in “Unsloth Dynamic 2.0 GGUFs”
Quality dips at lower bits
“They seem to make less errors at lower "average" quantizations than the unsloth 4 bit quants.”
Comment from 19 Jan 2026 in “GLM-4.7-Flash” “4 bit unsloth quants are good if you never ask for more than 20k context”
Comment from 16 Jun 2026 in “Running local models is good now” “They definitely feel a bit dazed.”
Comment from 28 Feb 2026 in “How do I cancel my ChatGPT subscription?”
What people recommend instead
The same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Mixed | Criticize | Approval | Months | Had for | Badge |
|---|---|---|---|---|---|---|---|
| Unsloth GGUF quantsThis page | 33 | 2 | 10 | 77% | 10 | — | Approved |
| Unsloth Dynamic GGUFs | 13 | 2 | 7 | 65% | 4 | — | — |
What people said
40 of the 49 opinions about it, picked in proportion to the three counts, newest first. Each links to the original comment.
“amazed by the Unsloth team by how much they can reduce the requirements”
“unsloth quants are amazing”
“It’s very possible using unsloth 2bit and 3bit quants”
“I get subjectively better results with Unsloth's 3 bit quants (UD-Q3-XL something), than their 4 bit quants”
“GGUF unsloth is so far the best for me.”
“errors out in LM Studio using the Unsloth quants”
“as they tend to be the best”
“I usually use Unsloth's GGUFs via llama.cpp”
“GGUFs from Unsloth can run on pretty much every single potato”
“the QAT is the business, indistinguishable from the full-fat version”
“I have had great luck with unsloth quants so far.”
“They are the most reliable in my experience”
“they make it first, then they run a lot of tests, KD, perplexity, etc, they publish it”
“I had several issues with unsloth gguf, even for models released a few months back like gemma 4”
“in my experience Unsloth ones are excellent”
“I've been having success with: https://unsloth.ai/docs/models/qwen3.6”
“Running unsloth versions with llama.cpp, constant issues.”
“Unsloth quants fix this in their Gemma releases.”
“I’m using Unsloth’s Q8_K_XL quant.”
“4 bit unsloth quants are good if you never ask for more than 20k context”
“Unsloth's quants are better than the original Google's QAT as posted in the article.”
“I tried Unsloth but got some weird problems”
“the unsloth versions can run on a potato and are actually useful”
“Unsloth are doing a fantastic job of publishing quants of the major models quickly”
“really love what unsloth is doing in this space”
“I had similar issue until I used Unsloth’s. I would recommend at least UD_6.”
“Lemonade builds of llama.cpp with Unsloth or Bartowski quants have proven to be excellent.”
“They are higher quality than the quants you can make yourself due to their imatrix datasets”
“something like Unsloth's UD q8 XL or q6 XL quants free up a lot of memory and bandwidth”
“we're now the best in terms of KLD and disk space”
“Just use the unsloth quants, they're great.”
“gave this guide to my Claude and he oneshot the unsloth and gemma4 set up on the old macbook”
“unsloth produces nice quants for it and even provides guidelines”
“Sometimes unsloth has broken ones for a particular model, sometimes no quants at all”
“Yes, it fails too. I’m using the unsloth q4_km quant.”
“The unsloth quants seem to have been fixed, so they are probably the go-to again”
“more like bugfixes for their broken first batch”
“They definitely feel a bit dazed.”
“the unsloth quants are great”
“They seem to make less errors at lower "average" quantizations than the unsloth 4 bit quants.”
Counts come from 1,791,911 Hacker News comments labeled by Claude, November 2025 to October 2026, with data through October 31, 2026. Counts updated 2 Nov 2026. How we count. Report a mistake.