base_model: orcarouter/Qwen3.8-Flash-Next-Uncensored
base_model_relation: quantized
library_name: mlx
pipeline_tag: image-text-to-text
license: other
license_name: qwen-community-1.0
license_link: LICENSE
tags:
- omlx
- mlx
- qwen4
- mtp
- vision
- tool-calling
Qwen3.8 Flash Next Uncensored oQ4e for oMLX
This is an oMLX-native quantization oforcarouter/Qwen3.8-Flash-Next-Uncensored.
It was built from pinned revisionfdf5fe3fd994c0cd6023c701d28f8820054f5aae.
It targets a 128 GiB Apple Silicon Mac and preserves the Qwen4 vision tower,
the checkpoint's own MTP head, and the checkpoint's native 262,144-token
architecture setting.
The artifact is intended for current oMLX Qwen4 support.
It is not a separate mlx-vlm release or compatibility target.
Quantization
The default layout is oQ4e affine quantization with group size 64.
The large PLE table uses four bits with group size 32 so oMLX can use its
SSD-backed Qwen4 PLE path.
Strict oQe calibration covered 937 modules over 1,024 sequences of 512 tokens.
Seven routed experts in base layers 0 and 1 were not selected by that fixed
calibration corpus.
All three routed projections in both affected blocks are therefore Q8 and do
not use invented importance values.
On the exact affected BF16 rows, Q8 reduced normalized reconstruction error by
about 164 times compared with Q4.
The base language-model head, both base hyper-connection mixer projections,
and the Qwen4 MTP embedding and hidden fusion projections remain BF16.
These additions favor output quality and MTP draft acceptance over the last
few GiB of compression.
The weights occupy 108,816,636,376 bytes across 22 safetensors shards.
oMLX setup
Add the downloaded model folder to oMLX and use a 100,000-token context limit
on a 128 GiB Mac.
Keep Qwen4 PLE SSD offload enabled for the full context window.
MTP is included, but keep it disabled for Swival.
The matched tool soak was less accurate with MTP enabled.
The certified Swival result requires oMLX 0.6.3 through this workspace'sscripts/serve launcher.
That launcher installs the tested tool-parser and total-context guards.
Bare oMLX 0.6.3 still mishandles some fenced tool examples and untyped native
Qwen XML arguments, so it is not the certified tool-calling configuration.
For Swival, set reasoning_effort at the profile level and in the provider
request body so a global setting cannot override the model profile:
reasoning_effort = "medium"
extra_body = { reasoning_effort = "medium" }
Do not enable a prompt cache for exact full-window certification.
The model's prompt plus requested output must remain within 100,000 tokens.
Why 100,000 tokens
The checkpoint retains its native 262,144-token architecture setting, but that
number is not a safe operating limit for this 128 GiB machine.
The advertised 100,000-token limit is the largest round target that passed both
exact API accounting and full-window retrieval with useful memory headroom.
At 100,000 tokens, the retrieval run peaked at 84.62 GiB of MLX memory and
91.01 GiB of physical memory, retained about 12.98 GiB of approximate available
memory, and caused no swap growth.
The 131,072-token attempt reached 118,784 processed tokens before oMLX predicted
a 98.02 GB peak against its 96.77 GB balanced prefill safety cap.
It could only fall back to 106,496 tokens, which was too close to the required
memory reserve for a reliable advertised limit.
Rounding down to 100,000 leaves room for oMLX, Swival, and normal operating
variance without changing the model's native positional metadata.
Validation
The artifact has passed structural verification of every shard and tensor,
the Q8 and BF16 safety policy, complete vision and MTP inventories, and the
native context configuration.SHA256SUMS binds all 22 final weight shards after the final directory rename.
Fresh oMLX 0.6.3 testing on an M5 Max with 128 GiB passed an exact 100,000-token
API budget.
A 99,999-token uncached prompt plus one output token completed successfully,
while both one-token-over requests were rejected.
Peak MLX allocation was 84.62 GiB, peak physical memory was 90.82 GiB, and swap
did not grow.
The long-context retrieval test also passed with a 99,744-token prompt and a
256-token output reserve.
It recovered exact codes near token positions 569, 51,721, and 98,256, finished
after 184 output tokens, peaked at 91.01 GiB of physical memory, and caused no
swap growth from its stable baseline.
The 131,072-token target is not supported by this safety profile.
It reached 118,784 processed tokens before the balanced guard predicted a
98.02 GB peak against its 96.77 GB prefill cap.
The real-image test identified Half Dome correctly.
MTP is operational and accepted 920 of 1,140 drafted tokens in the matched
greedy suite, but exact visible parity held for only one of eight prompts.
The final 60-case Swival certification meets the excellent tool-calling
requirement with MTP disabled.
All 50 positive cases passed on the first attempt and finally, and all 10
adversarial guard cases were protected.
This result uses the workspace oMLX parser guard and the tested Swival source
fixes for textual-call recovery and non-idempotent duplicate suppression.
MTP remains operational, but the matched MTP-on matrix scored 48 of 50 positive
cases and 10 of 10 adversarial guards.
It accepted 9,765 of 11,150 drafted tokens across 139 model requests, but the
two exact tool-argument regressions fail the no-regression requirement.
Keep MTP disabled for Swival.
License
The pinned source repository contains the Qwen Community License 1.0, which is
preserved in LICENSE and includes additional commercial terms.
The source model card labels the repository Apache-2.0, but that label does not
match the license file in the pinned source revision.
Review the included license before using or distributing this artifact.