alt Hacker NewsQwen/Qwen3.8-2.4T-A95B
Comments
https://unsloth.ai/docs/models/qwen3.8
The 1bit quant model is at an astonishing 397GB with 95B active per MOE. This literally puts Opus 4.5 performance level into a machine a normal person could buy, and still gets usable tokens/second.
The full lossless model BF16 is clocking at 4.9TB. The model card claims the model to be between Opus 4.8 and Fable 5. Again that's astonishing as getting a machine with 7TB RAM (with context + KV cache) is still within the realm of medium size companies.
Bad things: The open source version has its vision capability removed, and the context capped at 250k . I expect someone to bolt a Kimi 2.6 vision tower to it to restore the vision capability (at less performance of course). For context, I played around with extending the context to 600k for Qwen 3.5 397b, and the context remained stable up to around 480k. It'd be interesting to see if the same can be done to Q3.8 .
Also no out of the box DSpark/DFlash support. MTP is present so we should at least get some boost in TP speed.
Is this the largest ever open weight model release by parameter count? I think it is.
The card looks almost too good to be true
A ~5TB model.
Supposedly this is a Kimi k3 rival. Bit of a chonker, especially since they only released bf16 and fp8. So at launch this will be harder to serve than k3. No QAT on q4 means that someone with deep pockets (nvda?) will have to quant it, with plenty of calibration data. Should bring it ~1.3TB, so around k3 size.
License pretty similar to k3 with some caveats. Free to use for internal or <50M$ revenue / year. Limitations above that threshold for serving the model or services targeting coding / productivity agents.
Benchmarks are looking good, trading blows w/ opus4.8 and sol, generally 10-20p under fable. But that's neither here nor there w/ qwen, their benchmark to real world usage correlation has been iffy in the past.
The local model 3.8-27B announced for Friday, same time so ~48 hours from now. That'll be a bit more exciting for a lot more people, since 3.6 was quite good for local inference, and their 3.7-max -> 3.8-max shows a lot of improvement.