← back to catalog · registered 2026-08-22 13:56

Blackfrost-AI/Muse-Glimmer-30B-Abliterated-MLX-3bit-AWQ

Blackfrost-AI 30B multimodal second-order
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/Blackfrost-AI%2FMuse-Glimmer-30B-Abliterated-MLX-3bit-AWQ"
Response includes
  • classification m1
  • files 15
  • author_summary 19 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M1
Primary method

Direct removal

No other method signals detected in this model.
Confidence
MEDIUM
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=0 (base model)
  • no specific method indicators - defaulting to M1 (most common)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
Downloads · 30-day
160
Likes
0
Model age
2mo ago
created 2026-08-10
Downloads over time
Now160→from0↑0%
0591171760 on Aug 12160 on Aug 27160 on Aug 26Aug
Aug 12 → Aug 27 · 4 snapshots · spans 15 days

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Variants by this author 2 formats · 409 downloads combined

The same weights this author released in different packaging. Pick the format that matches your runtime.

Metadata

License
apache-2.0
Tags
mlx safetensors muse_glimmer muse-glimmer abliterated awq quantized apple-silicon 3-bit image-text-to-text conversational base_model:Blackfrost-AI/Muse-Glimmer-30B-Abliterated-BF16

Related

Total size
14.8 GB
Files
15
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-08-11 02:34

Files by quantization

Auxiliary files 15 files 14.9 GB
model-00002-of-00003.safetensors 4.98 GB 032787f0 download
model-00001-of-00003.safetensors 4.98 GB 5687357d download
model-00003-of-00003.safetensors 4.87 GB c00a6fc3 download
tokenizer.json 26.8 MB ad80c0cb download
model.safetensors.index.json 213 KB 20f9f0c4 download
tokenizer_config.json 78.1 KB d1b80588 download
LICENSE 11.1 KB d6456956 download
chat_template.jinja 9.31 KB 6fbdf45f download
muse_kratos_template.jinja 9.31 KB 6fbdf45f download
config.json 6.64 KB f38bc941 download
USAGE_POLICY.md 5.11 KB 1a9ed6cf download
README.md 2.81 KB ba13a051 download
.gitattributes 1.53 KB 52373fe2 download
processor_config.json 395 B 82e0b73b download
generation_config.json 148 B 3495c31e download

README current version from Hugging Face


license: apache-2.0
base_model:

  • Blackfrost-Research/Muse-Glimmer-30B-Abliterated-BF16
    tags:
  • muse-glimmer
  • abliterated
  • mlx
  • awq
  • quantized
  • apple-silicon
  • 3-bit
    pipeline_tag: image-text-to-text
    library_name: mlx

MUSE-GLIMMER-30B-ABLITERATED-MLX-3bit-AWQ

3-bit AWQ MLX for Apple Silicon — activation-aware, quality-recovered 3-bit

Built by Blackfrost · Las Vegas, NV

Refusal benchmark — R1-HARMFUL-BENCH-450

Measured on the abliterated model — quantization holds it, no refusal snapback:

Metric Result
True refusal (harmful, n=300) 0 / 300 = 0.0%
True refusal (full 450) 0 / 450 = 0.0%
Substring-harmful 0 / 300
Substring-all 2 / 450 (XSTest false positives)
Errors 0

Why this model exists

Muse Glimmer is Meta Superintelligence Labs' 30B agentic, on-device model. This is the abliterated build — refusal behavior removed via a Blackfrost weight-change process — packaged as MLX 3-bit AWQ — ~15 GB for Apple-silicon Macs. The local footprint is the product.


Specifications

Architecture muse_glimmer — dense, 52 layers, hidden 6656, GQA (32 q / 2 kv), sliding-window attention, + vision tower
Base meta-models/Muse-Glimmer-30B — Meta, Apache-2.0
Transform Abliterated — refusal behavior removed via a Blackfrost weight-change process; multimodal capability intact
Format MLX 3-bit AWQ — ~15 GB
Context 131,072

Serving (Apple Silicon / MLX)

pip install mlx-lm
# one-off generate:
mlx_lm.generate --model Blackfrost-Research/Muse-Glimmer-30B-Abliterated-MLX-3bit-AWQ --prompt "Write a binary search in Python." --max-tokens 1024
# OpenAI-compatible server:
mlx_lm.server --model Blackfrost-Research/Muse-Glimmer-30B-Abliterated-MLX-3bit-AWQ --port 8080

Or open it directly in LM Studio (MLX runtime) on an Apple-silicon Mac.

Sampling (Meta): temperature 1.0, top_p 0.95, top_k 64. It's a heavy thinker — use a generous max_tokens (≥ 1024) and steer depth with a Reasoning strength: low/medium/high/xhigh system line. Reasoning is returned separately from the final answer.


Built by Blackfrost · Las Vegas, NV. Not affiliated with Meta.

Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration