Qwen3.8 Flash Next

IQ4_XS · mradermacher · Measured configuration

84.19 / 100 Mango weighted score · 35.33 min matched workload · 12/12 mechanical agent passes

The mradermacher IQ4_XS packaging of OrcaRouter's refusal-modified Qwen3.8 Flash Next. This mirror preserves the exact single-file quant used in the original Mango comparison.

Benchmark results · Files · Download · Runtime · Provenance · Full collection

Join Infinity on Discord

Get model support, request models and quants, discuss benchmarks, and follow new releases and model improvements.

Join the Infinity community

All model tests, per-question scores, experiment history and scoring methods

Qwen effort study — separate full-404 cohort: medium 82.87/100 versus xhigh 84.41/100; paired difference +1.54 points (95% interval [-1.71, +4.82]). 404 collected questions, 403 valid questions graded per effort; one invalid reference annotation excluded symmetrically. The historical 202-question cross-model comparison below is unchanged. Study, timing and blind subjective results.

Preserved and documented by isilicon / Mango. The original model, modification and quantization remain the upstream creators' work. This repository adds attributed preservation and exact-configuration evaluation, with the selected source bytes unchanged.

At a glance

GB is decimal and GiB is binary. Payload size is disk/storage size, not a claim that the model fits in the same amount of VRAM. Runtime, KV cache, working memory and companions add their own requirements.

Mango evaluation

Highest overall point estimate in the ten-configuration cohort, with 19/20 local coding tests and 31/40 local function-call checks passing. All 12 agent cases satisfy the mechanical completion/evidence/permissions checks.

The matched workload took 35.33 minutes, versus 23.98 for Qwen27 MLX4. Its quality advantage over Qwen remains uncertain on this small sample; the paired intervals below are more informative than the rank alone.

Domain scores equally average their component task means. The overall score applies the weights above. EQ is continuous; the 12-case mechanical agent pass rate is only one component of the agentic domain.

Task-level quality and uncertainty

These are small released subsets and local graders. HumanEval/BFCL scores are not the complete official benchmark pipelines. Agent checks cover completion, evidence, permissions and applicable pricing totals; other answer semantics remain unjudged.

Recorded performance

Workload time excludes loading, restarts, wipe downtime and idle gaps. Latency/TTFT summaries cover suite requests only. The reported token rate divides engine usage counts by request time; it is not native decode throughput and token accounting differs across engines. No concurrent-serving result is implied.

Matched cohort comparison

The old mixed GLM row is a historical, internally capped baseline whose exact weights are unavailable. Other rows also differ in quant, engine, context and effort. Point ranks do not settle the quality ordering.

Paired score differences

Intervals use 10,000 paired task-stratified bootstrap resamples, seed 41, with fixed task/domain weights. An interval containing zero does not establish a quality winner. See methodology for scope and definitions.

Download the current full numerical report and methodology. The original compact evaluation is retained as the earlier preservation snapshot; its model score is unchanged, while the current report includes the expanded cohort. The numerical report contains aggregate statistics and hashes; no prompts or generated answers are published.

Runtime and compatibility

Use a llama.cpp build that understands qwen4_exp. Older bundled runtimes can reject this architecture. The Mango run used llama.cpp on an NVIDIA Blackwell GPU with CPU/RAM offload as required; this is not an Apple Silicon measurement.

32,768-token configured benchmark window; medium effort after the proxy's high-to-medium mapping.

The local evaluation covers text/tool tasks. It does not establish the full advertised context ceiling, image/audio quality or every optional speculative-decoding path. Publisher capabilities and Mango-verified behavior are separate scopes.

Companion artifacts

No separate companion mirror is declared in this selected manifest. This does not imply that every optional upstream projector, draft head or multimodal path is included; consult the exact file list.

Files and integrity

Full paths, byte sizes and complete SHA-256/Git blob identities are in SOURCE_MANIFEST.json. Available original README/license/notice files are also preserved. MIRROR_STATUS.json records the earlier source-file verification; MODEL_CARD_STATUS.json records this documentation refresh separately.

Download and use

The commands pin the already-verified mirror weight revision. Documentation can advance on main without changing that preserved weight snapshot. Keep original subfolders and all shards.

python -m pip install -U huggingface_hub

hf download isilicon/mango-qwen38-flashnext-iq4xs \
  Qwen3.8-Flash-Next-Uncensored.IQ4_XS.gguf \
  --revision 13bc0ca37ef05ec74c91b61b26fb052efac421c0 \
  --local-dir ./weights/qwen38-flashnext-iq4xs

Load example and setup scope

llama-server \
  --model ./weights/qwen38-flashnext-iq4xs/Qwen3.8-Flash-Next-Uncensored.IQ4_XS.gguf \
  --host 127.0.0.1 --port 8080 --ctx-size 32768 \
  --jinja --flash-attn on --cache-type-k q8_0 --cache-type-v q8_0

This is a minimal text-serving example for a compatible build, not a replay of every benchmark fitting/offload flag. The recorded configuration and engine caveats define the tested scope.

Provenance and license

OrcaRouter supplied the modified checkpoint; mradermacher supplied this GGUF quantization. It is a different package from the three-shard direct Orca IQ4_XS mirror. Both derive from the Orca lineage, so comparing them does not isolate the effect of refusal removal.

Original notices, base-model conditions and the publisher's stated limitations still apply. Downloadable preservation does not add new license rights. Source-described refusal modification and publisher-side benchmark claims have not been independently re-measured by this Mango task cohort.

Original publisher documentation

The unchanged upstream README and pinned source card retain the original usage guidance, technical discussion, attribution, caveats and citations.

Read the complete original publisher card

Source reference. The following is the original publisher's documentation. Its other quants, full-model files, context claims and benchmark tables can exceed the selection in this mirror. Relative asset/file URLs are qualified to the pinned original source; publisher measurements remain separate from Mango measurements.

About

static quants of https://huggingface.co/orcarouter/Qwen3.8-Flash-Next-Uncensored

For a convenient overview and download list, visit our model page for this model.

weighted/imatrix quants are available at https://huggingface.co/mradermacher/Qwen3.8-Flash-Next-Uncensored-i1-GGUF

Usage

If you are unsure how to use GGUF files, refer to one of TheBloke's READMEs for more details, including on how to concatenate multi-part files.

Provided Quants

(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)

Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better):

And here are Artefact2's thoughts on the matter: https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9

FAQ / Model Request

See https://huggingface.co/mradermacher/model_requests for some answers to questions you might have and/or if you want some other model quantized.

Thanks

I thank my company, nethype GmbH, for letting me use its servers and providing upgrades to my workstation to enable this work in my free time. Additional thanks to @nicoboss for giving me access to his private supercomputer, enabling me to provide many more imatrix quants, at much higher quality, than I would otherwise be able to.

Preservation record

Selected source bytes are verified against pinned metadata; the original eight graded mirrors also passed the previously recorded anonymous 1 MiB weight-prefix checks. This is not a full independent-device restore/load test. Hugging Face repository copies share a provider and do not guarantee permanent storage.

Collection · Source manifest · Original documentation