what's this?
what happend?
Oh this is our investigation into Qwen3.5-35B-A3B-GGUF quant levels and stuff :) We're announcing results later today!
Oh this is our investigation into Qwen3.5-35B-A3B-GGUF quant levels and stuff :) We're announcing results later today!
I suppose so, but why do you need to upload all of them? I don't think we can do anything with them?
These are research artifacts for people to verify results if they wish!
@danielhanchen Thanks for your efforts man! Plz post on LinkedIn or your blogbost once you get the final new fully optimized GGUFs so we can jump right in!
Yep I will update everyone soon!
These are research artifacts for people to verify results if they wish!
ok thanks for your work :)
average Daniel W 🔥
Hey guys, I'm sure most of you know already but if you didn't, here's the update: https://x.com/UnslothAI/status/2027449469596545535
"Qwen3.5 is now updated with improved tool-calling & coding performance! See improvements via Claude Code, Codex.
We also benchmarked GGUFs & removed MXFP4 layers from 3 quants.
We will be re-uploading the non UD ones in a few hours with the chat template fixes if all goes well."
Thanks so much :)
Analysis + MXFP4 investigation: https://unsloth.ai/docs/models/qwen3.5/gguf-benchmarks
CC: @QuackityDuckity @engrtipusultan @lightenup @dobriak @eleius @Hugy-Bear @itsthenewmeta @felipedpm @watchingyousleep @iliasslasri @Simon716 @Poro7 @dzupin @ykarout @LaikaFramework @lly0571 @balukumar @naumf @MaCgl @Nimbz @dehnhaide @tomngdev @balukumar @icewall905 @latent-variable @gkubon @sudo-0x2a
Thanks for your exceptional diligence & openness!!
Good Job.. 👍