license: apache-2.0 base_model: AEON-7/Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-BF16 language:

• en
• zh
• multilingual library_name: transformers pipeline_tag: text-generation tags:
• 27b
• official-release
• abliterated
• abliterix
• aeon
• aeon-7
• agentic
• bf16
• blackwell
• chat
• chunked-prefill
• coding
• conversational
• dgx-spark
• dflash
• fp4
• fp8
• function-calling
• gated-deltanet
• gb10
• gdn
• hybrid-attention
• image-text-to-text
• instruct
• linear-attention
• long-context
• mamba
• mixed-precision
• modelopt
• mtp
• multimodal
• nvidia-modelopt
• nvfp4
• openai-compatible
• quantized
• qwen
• qwen3
• qwen3.5
• qwen3.8
• qwen3_5
• reasoning
• refusal-removed
• rtx-5090
• rtx-pro-6000
• sm_120
• sm_121a
• speculative-decoding
• thinking
• tool-calling
• uncensored
• unfiltered
• vision
• vision-language
• vllm

Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-NVFP4-MIXED

AEON Uncensored NVFP4 Mixed Precision Release

Announcement. This Hub card is the release notice for the MIXED deploy cut. Deeper release notes and expanded serve recipes land on Patreon as more battle testing and raw data comes in — not all at once on day one.

I've spent a lot of months learning what abliteration actually costs. Turns out the trap most people fall into is chasing a vanity KL of zero and calling every judge-R a leftover refusal. That is how you lobotomize a 27B. I wanted the opposite: willfully compliant, coherent, useful — and still a real model when the answer gets long.

So this is the ~23G ModelOpt mixed deploy cut of AEON-7/Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-BF16. Not a blunt NVFP4 dump. Not a dequant-edit-requant circus. A lattice designed so the last layers can still write code that passes unit tests, while the bulk of the MLP rides NVFP4 where Blackwell is actually hungry for it.

Official MIXED deploy cut. Cosmic remains a later brand for the intelligence heal. Do not treat a long-context loop as the last word on this line.

Model family (pick your seat)

Same will. Different seats. The BF16 is the source of truth. This card is what you actually serve day to day.

The lattice (why MIXED, not "just NVFP4")

AEON MIXED quant lattice — NVFP4 body, FP8 writers, last-8 full MLP FP8, BF16 vision/MTP

I do not trust equal-bitwidth recipes on a hybrid GDN / full-attn stack. The first Mix B bake FP8'd only the last-8 down_proj and left gate/up on NVFP4. Coding cratered to 0.694 on AEON Bench unit tests — real syntax and assert fails, not a moody judge. Leftover-KL on those miss prefixes lit up layers 56–63. So the sibling bake did the honest thing: last-8 full MLP (gate + up + down) → FP8. Coding jumped to 0.833.

Calib 1024×2048. Export gate: found=168 missing=0 last8_mlp_fp8=24. Parent BF16 untouched. Post-export: grafted unmodified mtp.* (15 BF16 tensors, 0.791G) from Ultimate BF16 — ModelOpt had dropped the head.

What you get

• Uncensored will — hall monitor gone; answers instead of sermons (same lineage as the BF16 master)
• Coding that ships — AEON Bench Coding 0.833 vs prior Mix B 0.694 (and ahead of stock Unsloth NVFP4 Coding 0.806 on the public card)
• Agentic legs — Hermes 0.918, OpenCode 0.818 on the same local follow-on (OpenClaw still the soft spot)
• Deployable size — ~23.8G (NVFP4 body + BF16 vision + BF16 MTP) on one Spark / one 5090 / one PRO 6000
• Full multimodal + MTP — vision tower and video preprocessor intact; native MTP head for RTX speculative decode
• Honest scoreboard — Intelligence (Math/Reasoning/Prose) still trails stock Qwen's 84.5 composite; Cosmic is the heal track, not this card pretending otherwise

Local AEON Suite v4 run ceaac9e418 (GB10, think 32k / max 65k / retry 131k, c=8, mothership fail-closed):

Attested mothership Global (aeon-bench.com, same MIXED tree): overall ~91.9, Perf dial 100, peak_agg 237.67 tok/s (wave-peak, Coding@c16) on the single Spark Dynamic DFlash lattice seat below.

QuickStart — Spark Dynamic DFlash lattice + RTX seats

Use ghcr.io/aeon-7/aeon-vllm-ultimate:2026-09-11-v0.29.0-omni on DGX Spark / GB10 for the new engine tip (upstream vLLM 0.29.0, sm_121a bake, digest sha256:2421bb1228a85370c1c50adb31f605c4361acf4d48d65282fcb919e74f34fae7). It keeps the AEON carries (DFlash/DFlash2, reasoning-EOS force-end, NVFP4 paths) and is the cut where dual-Spark TP=2 stays coherent if you set the B0 knobs: VLLM_ALLREDUCE_USE_FLASHINFER=0, fuse_allreduce_rms=false, --disable-custom-all-reduce, and NCCL_IB_HCA=<dev>:1 with a single equals. TP=1 A/B vs 2026-09-07-reasoning-eos held quality and was slightly faster. MIXED still needs the ModelOpt #54367 bind on every rank — in-tree 0.29 ModelOpt alone does not load this lattice. Pin the dated tag until you confirm GHCR :latest shares that digest; rollback remains :2026-09-07-reasoning-eos.
Use ghcr.io/aeon-7/aeon-vllm-ultimate-rtx:latest on RTX 5090 and RTX PRO 6000 (sm_120; promoted from 2026-09-07-omni-mm = ModelOpt #54367 plus Spark-parity vllm-omni multimodal/omni stack). Pin dated RTX tags if you want known-good: 2026-09-07-omni-mm (or 2026-09-07-modelopt for #54367-only). Do not cross Spark and RTX images. Spark is sm_121a / aarch64 UMA. The RTX image is sm_120 / amd64 dedicated VRAM.

Vision / video / audio: RTX :latest / :2026-09-07-omni-mm registers the multimodal processor for Qwen3_5ForConditionalGeneration (same family as Spark omni) and layers vllm-omni==0.27.0rc1 for Omni speech-out models. Keep --limit-mm-per-prompt as in the recipes below. Re-pull: docker pull ghcr.io/aeon-7/aeon-vllm-ultimate-rtx:latest.

Stale Aug-21 RTX :latest symptom: 'MergedColumnParallelLinear' object has no attribute 'data' during fused gate_up modelopt_mixed load — bind-mount modelopt-54367.py until you pull the 2026-09-07 image.

Gen defaults for all seats (Qwen3.8 card): temperature 0.6, top_p 0.95, top_k 20, repetition_penalty 1.0. Thinking on: chat_template_kwargs={"enable_thinking": true, "reasoning_effort": "medium"}.

Prefer repetition_penalty: 1.0 for tool calling / structured output. Values above 1.0 can penalize repeated structural tokens (e.g. /parameter, XML tags) and break parsers.

ModelOpt hard rules (this MIXED tree)

Weights mount (HF snapshot symlink trees)

Hugging Face Hub snapshot dirs under ~/.cache/huggingface/hub/models--…/snapshots/<rev> are symlink trees into blobs/. Mounting only the snapshot path into Docker often yields dangling symlinks inside the container. Prefer one of:

1. Mount the HF cache root and serve by repo-id + revision (recommended): -v $HOME/.cache/huggingface:/root/.cache/huggingface:ro then vllm serve AEON-7/Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-NVFP4-MIXED --revision <rev> …
2. Or mount a fully materialized copy (real files, no dangling links), e.g. a local huggingface-cli download … --local-dir tree.

Do not mount only …/snapshots/<rev> unless you have verified every weight path resolves inside the container.

Knob cheat-sheet

--gpu-memory-utilization note (Spark): default 0.80. That can be excessive when the box also runs secondary services — downshift util to leave headroom. On a dedicated LLM-only Spark, pushing to 0.85 may gain slight performance. Going beyond 0.80 risks OOM under high KV-cache pressure, so 0.85 is optional dedicated-only, not the published default.

[!TIP]

Dynamic DFlash lattice (single Spark seat)

This is the locked peak-16 map — the one published single-Spark recipe (quality and throughput).

• Native key: num_speculative_tokens_per_batch_size
• Exact map JSON: [[1,1,10],[2,2,10],[3,4,8],[5,8,7],[9,10,6],[11,12,5],[13,14,4],[15,16,3]]
• Runtime K: c1–2→10, c3–4→8, c5–8→7, c9–10→6, c11–12→5, c13–14→4, c15–16→3
• Peak measured: 237.67 tok/s wave-peak Coding@c16
• Default util 0.80 (downshift for secondary services; optional 0.85 dedicated-only — beyond 0.80 risks OOM with high KV)
• Full docker block: One DGX Spark — Dynamic DFlash lattice

1) One DGX Spark — Dynamic DFlash lattice (quality + throughput)

One published single-Spark serve seat. Locked Dynamic DFlash lattice via num_speculative_tokens_per_batch_size, MRv2 + FULL_AND_PIECEWISE, default util 0.80, seqs 16, 262k, KV fp8, chunked prefill, no prefix cache, TRITON_ATTN. Image pin 2026-09-11-v0.29.0-omni (digest sha256:2421bb…); rollback :2026-09-07-reasoning-eos. #54367 bind required on 0.29 for MIXED.

This seat is both quality and throughput — not a demoted “Perf-only” lane.

Measured AEON Bench peak overall (wave-peak concurrent window, not decode×conc): 237.67 tok/s Coding@c16.

Util: default 0.80. Downshift when co-running secondary services. Optional 0.85 on a dedicated LLM-only box for a slight gain — going beyond 0.80 risks OOM with high KV-cache usage.

MODEL=$HOME/.cache/huggingface/hub/models--AEON-7--Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-NVFP4-MIXED/snapshots/<rev>
DRAFT=/path/to/z-lab__Qwen3.8-27B-DFlash2
PATCH=/path/to/modelopt-54367.py   # required on 2026-09-11-v0.29.0-omni for MIXED; optional on :latest / 2026-09-07-reasoning-eos if already folded
IMAGE=ghcr.io/aeon-7/aeon-vllm-ultimate:2026-09-11-v0.29.0-omni   # pin dated until :latest digest matches 2421bb…; rollback :2026-09-07-reasoning-eos

docker rm -f aeon-mixed-spark 2>/dev/null
docker run -d --gpus all --network host \
  --name aeon-mixed-spark \
  -e VLLM_USE_V2_MODEL_RUNNER=1 \
  -e VLLM_ENABLE_CUDA_COMPATIBILITY=0 \
  -e VLLM_ALLOW_LONG_MAX_MODEL_LEN=1 \
  -v "$MODEL:/model:ro" -v "$DRAFT:/draft:ro" \
  -v "$PATCH:/usr/local/lib/python3.12/site-packages/vllm/model_executor/layers/quantization/modelopt.py:ro" \
  --entrypoint vllm "$IMAGE" serve /model \
  --served-model-name aeon \
  --host 0.0.0.0 --port 8000 \
  --gpu-memory-utilization 0.80 \
  --max-model-len 262144 \
  --max-num-seqs 16 \
  --max-num-batched-tokens 16384 \
  --kv-cache-dtype fp8 \
  --enable-chunked-prefill \
  --no-enable-prefix-caching \
  --tool-call-parser qwen3_coder \
  --enable-auto-tool-choice \
  --reasoning-parser qwen3 \
  --limit-mm-per-prompt '{"image":4,"video":2}' \
  --attention-backend TRITON_ATTN \
  --compilation-config '{"cudagraph_mode":"FULL_AND_PIECEWISE"}' \
  --trust-remote-code \
  --speculative-config '{"method":"dflash","model":"/draft","num_speculative_tokens":10,"num_speculative_tokens_per_batch_size":[[1,1,10],[2,2,10],[3,4,8],[5,8,7],[9,10,6],[11,12,5],[13,14,4],[15,16,3]],"attention_backend":"TRITON_ATTN"}' \
  --override-generation-config '{"temperature":0.6,"top_p":0.95,"top_k":20,"min_p":0.0,"presence_penalty":0.0,"repetition_penalty":1.0}'

Ready when /v1/models lists aeon. Smoke: chat completions, model=aeon, short prompt.

Runtime K by batch: c1–2→10, c3–4→8, c5–8→7, c9–10→6, c11–12→5, c13–14→4, c15–16→3.

2) Two DGX Sparks — TP=2 quality (1M YaRN + DFlash2 n=7)

Use only when two Sparks share a fast link (InfiniBand or RoCE). Each box needs the same $MODEL, $DRAFT, $PATCH, and image. Start rank 1 (headless) first, then rank 0 (API). Clients use http://$MASTER_IP:8000.

MODEL=$HOME/.cache/huggingface/hub/models--AEON-7--Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-NVFP4-MIXED/snapshots/<rev>
DRAFT=/path/to/z-lab__Qwen3.8-27B-DFlash2
PATCH=/path/to/modelopt-54367.py
IMAGE=ghcr.io/aeon-7/aeon-vllm-ultimate:latest
MASTER_IP=10.10.10.1
IFACE=enp1s0f0np0
IB_HCA==rocep1s0f0:1
HF_OVERRIDES='{"text_config":{"rope_parameters":{"mrope_interleaved":true,"mrope_section":[11,11,10],"rope_type":"yarn","rope_theta":10000000,"partial_rotary_factor":0.25,"factor":4.0,"original_max_position_embeddings":262144}}}'

Change MASTER_IP, IFACE, and IB_HCA to this pair (ip -br link). The == in IB_HCA is required (NCCL exact match). No InfiniBand? Drop --device /dev/infiniband, -e NCCL_IB_HCA=..., -e NCCL_IB_GID_INDEX=3, and --cap-add IPC_LOCK; keep NCCL_SOCKET_IFNAME on the fast nic.

Do not pass --quantization. Bind-mount #54367. Set VLLM_USE_V2_MODEL_RUNNER=0.

Rank 1 — worker (headless)

On the second Spark:

NODE_IP=10.10.10.2
RANK=1
NAME=tp2-node1

docker rm -f "$NAME" 2>/dev/null
docker run -d --name "$NAME" --gpus all --ipc=host --shm-size=16g --net=host \
  -e VLLM_HOST_IP="$NODE_IP" \
  -e NCCL_SOCKET_IFNAME="$IFACE" -e GLOO_SOCKET_IFNAME="$IFACE" \
  -e NCCL_IB_HCA="$IB_HCA" -e NCCL_IB_GID_INDEX=3 -e NCCL_DEBUG=WARN \
  --device /dev/infiniband --cap-add IPC_LOCK --ulimit memlock=-1:-1 \
  -e VLLM_USE_FLASHINFER_SAMPLER=1 \
  -e PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True \
  -e VLLM_ALLOW_LONG_MAX_MODEL_LEN=1 \
  -e VLLM_USE_V2_MODEL_RUNNER=0 \
  -v "$MODEL":/model:ro -v "$DRAFT":/draft:ro \
  -v "$PATCH:/usr/local/lib/python3.12/site-packages/vllm/model_executor/layers/quantization/modelopt.py:ro" \
  --entrypoint vllm "$IMAGE" serve /model \
    --headless \
    --tensor-parallel-size 2 --nnodes 2 --node-rank 1 \
    --master-addr "$MASTER_IP" --master-port 29501 \
    --kv-cache-dtype fp8_e4m3 --attention-backend TRITON_ATTN \
    --max-model-len 1000000 --max-num-seqs 16 --max-num-batched-tokens 16384 \
    --gpu-memory-utilization 0.70 \
    --disable-custom-all-reduce \
    --enable-chunked-prefill --no-enable-prefix-caching \
    --mamba-cache-mode align \
    --limit-mm-per-prompt '{"image":4,"video":2}' \
    --speculative-config '{"method":"dflash","model":"/draft","num_speculative_tokens":7}' \
    --hf-overrides "$HF_OVERRIDES" \
    --override-generation-config '{"temperature":0.6,"top_p":0.95,"top_k":20,"min_p":0.0,"presence_penalty":0.0,"repetition_penalty":1.0}' \
    --trust-remote-code

Rank 1 must have --headless. Do not give it --host or --port.

Rank 0 — API

On the first Spark:

NODE_IP=10.10.10.1
RANK=0
NAME=tp2-node0

docker rm -f "$NAME" 2>/dev/null
docker run -d --name "$NAME" --gpus all --ipc=host --shm-size=16g --net=host \
  -e VLLM_HOST_IP="$NODE_IP" \
  -e NCCL_SOCKET_IFNAME="$IFACE" -e GLOO_SOCKET_IFNAME="$IFACE" \
  -e NCCL_IB_HCA="$IB_HCA" -e NCCL_IB_GID_INDEX=3 -e NCCL_DEBUG=WARN \
  --device /dev/infiniband --cap-add IPC_LOCK --ulimit memlock=-1:-1 \
  -e VLLM_USE_FLASHINFER_SAMPLER=1 \
  -e PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True \
  -e VLLM_ALLOW_LONG_MAX_MODEL_LEN=1 \
  -e VLLM_USE_V2_MODEL_RUNNER=0 \
  -v "$MODEL":/model:ro -v "$DRAFT":/draft:ro \
  -v "$PATCH:/usr/local/lib/python3.12/site-packages/vllm/model_executor/layers/quantization/modelopt.py:ro" \
  --entrypoint vllm "$IMAGE" serve /model \
    --served-model-name aeon --host 0.0.0.0 --port 8000 \
    --tensor-parallel-size 2 --nnodes 2 --node-rank 0 \
    --master-addr "$MASTER_IP" --master-port 29501 \
    --kv-cache-dtype fp8_e4m3 --attention-backend TRITON_ATTN \
    --max-model-len 1000000 --max-num-seqs 16 --max-num-batched-tokens 16384 \
    --gpu-memory-utilization 0.70 \
    --disable-custom-all-reduce \
    --enable-chunked-prefill --no-enable-prefix-caching \
    --mamba-cache-mode align \
    --limit-mm-per-prompt '{"image":4,"video":2}' \
    --speculative-config '{"method":"dflash","model":"/draft","num_speculative_tokens":7}' \
    --reasoning-parser qwen3 --tool-call-parser qwen3_coder --enable-auto-tool-choice \
    --hf-overrides "$HF_OVERRIDES" \
    --override-generation-config '{"temperature":0.6,"top_p":0.95,"top_k":20,"min_p":0.0,"presence_penalty":0.0,"repetition_penalty":1.0}' \
    --trust-remote-code

docker logs -f tp2-node0
curl -s http://127.0.0.1:8000/health

1M on TP=2 is the long-context recipe. It is slower than 64k.

TP=2 — speed (64k) deltas

Same run as above, but:

• remove -e VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
• remove --hf-overrides
• --max-model-len 65536
• --gpu-memory-utilization 0.60
• keep DFlash2 n=7, seqs 16, #54367 patch, no --quantization

3) RTX 5090 — aeon-vllm-ultimate-rtx:latest

32 GB dedicated. Long-context chat seat: TRITON_ATTN + fp8_e4m3 KV (FlashInfer + fp8 KV can garbage on sm_120 at long ctx). Seqs 4, util 0.92–0.95, 131k, no YaRN. Native MTP n=3 is in the checkpoint (~1% KV cost).

aeon-vllm-ultimate-rtx:latest / 2026-09-07-omni-mm folds #54367 and Spark-parity vllm-omni (vision/video/audio + Omni speech-out). 2026-09-07-modelopt is the #54367-only intermediate. On older Aug-21 images, set PATCH and bind-mount as below (harmless no-op if the file matches the folded copy).

# Prefer HF cache mount + repo-id (avoids snapshot symlink dangling links):
#   -v $HOME/.cache/huggingface:/root/.cache/huggingface:ro
#   serve AEON-7/Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-NVFP4-MIXED --revision <rev>
MODEL=$HOME/.cache/huggingface/hub/models--AEON-7--Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-NVFP4-MIXED/snapshots/<rev>
PATCH=/path/to/modelopt-54367.py   # optional on :latest / 2026-09-07-omni-mm; required on Aug-21
IMAGE=ghcr.io/aeon-7/aeon-vllm-ultimate-rtx:latest   # or :2026-09-07-omni-mm

docker run -d --name aeon-mixed-5090 --gpus all --ipc=host --shm-size=8g -p 8000:8000 \
  -e VLLM_ALLOW_LONG_MAX_MODEL_LEN=1 \
  -e PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True \
  -v "$MODEL:/model:ro" \
  -v "$PATCH:/usr/local/lib/python3.12/dist-packages/vllm/model_executor/layers/quantization/modelopt.py:ro" \
  --entrypoint vllm "$IMAGE" serve /model \
  --served-model-name aeon --host 0.0.0.0 --port 8000 \
  --max-model-len 131072 --max-num-seqs 4 --max-num-batched-tokens 8192 \
  --gpu-memory-utilization 0.92 --kv-cache-dtype fp8_e4m3 \
  --mamba-ssm-cache-dtype bfloat16 --attention-backend TRITON_ATTN \
  --enable-chunked-prefill --no-enable-prefix-caching \
  --limit-mm-per-prompt '{"image":2,"video":1}' \
  --speculative-config '{"method":"mtp","num_speculative_tokens":3,"attention_backend":"TRITON_ATTN"}' \
  --reasoning-parser qwen3 --tool-call-parser qwen3_coder --enable-auto-tool-choice \
  --override-generation-config '{"temperature":0.6,"top_p":0.95,"top_k":20,"min_p":0.0,"presence_penalty":0.0,"repetition_penalty":1.0}' \
  --trust-remote-code

If KV will not allocate at 131k, try --gpu-memory-utilization 0.95 then drop --max-num-seqs to 2. Do not add a fat DFlash drafter on 32 GB for the long seat.

4) RTX PRO 6000 (96 GB, sm_120) — aeon-vllm-ultimate-rtx:latest

Same image family as the 5090. More VRAM → raise concurrency and context headroom. Pin TRITON_ATTN for long context. If you use MTP, put "attention_backend":"TRITON_ATTN" inside --speculative-config (top-level flag does not always propagate to the drafter).

Same rule as the 5090: :latest / 2026-09-07-omni-mm folds #54367 + omni/MM; bind-mount modelopt-54367.py on Aug-21. Prefer HF cache + repo-id over snapshot-only mounts.

MODEL=$HOME/.cache/huggingface/hub/models--AEON-7--Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-NVFP4-MIXED/snapshots/<rev>
PATCH=/path/to/modelopt-54367.py   # optional on :latest / 2026-09-07-omni-mm; required on Aug-21
IMAGE=ghcr.io/aeon-7/aeon-vllm-ultimate-rtx:latest   # or :2026-09-07-omni-mm

docker run -d --name aeon-mixed-pro6000 --gpus all --ipc=host --shm-size=16g -p 8000:8000 \
  -e VLLM_ALLOW_LONG_MAX_MODEL_LEN=1 \
  -e PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True \
  -v "$MODEL:/model:ro" \
  -v "$PATCH:/usr/local/lib/python3.12/dist-packages/vllm/model_executor/layers/quantization/modelopt.py:ro" \
  --entrypoint vllm "$IMAGE" serve /model \
  --served-model-name aeon --host 0.0.0.0 --port 8000 \
  --max-model-len 262144 --max-num-seqs 8 --max-num-batched-tokens 8192 \
  --gpu-memory-utilization 0.80 --kv-cache-dtype fp8_e4m3 \
  --mamba-ssm-cache-dtype bfloat16 --attention-backend TRITON_ATTN \
  --enable-chunked-prefill --no-enable-prefix-caching \
  --limit-mm-per-prompt '{"image":4,"video":2}' \
  --reasoning-parser qwen3 --tool-call-parser qwen3_coder --enable-auto-tool-choice \
  --override-generation-config '{"temperature":0.6,"top_p":0.95,"top_k":20,"min_p":0.0,"presence_penalty":0.0,"repetition_penalty":1.0}' \
  --trust-remote-code

Native MTP (preferred on this seat):

--speculative-config '{"method":"mtp","num_speculative_tokens":3,"attention_backend":"TRITON_ATTN"}'

For Qwen long think, prefer a per-request thinking_token_budget over blindly raising max_tokens — otherwise the model can spend the whole budget inside <think>.

How this was built

AEON-7/Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-BF16
        ↓
  NVIDIA ModelOpt 0.46  ·  MSE + fp8_scale_sweep  ·  1024×2048
        ↓
  Mix B (last-8 down_proj FP8 only)  → Coding 0.694
        ↓
  leftover-KL on coding unit-test misses
        ↓
  this sibling: last-8 FULL MLP FP8
        ↓
  AEON Bench Coding 0.833

The unaligned edge

I figure if you own the hardware and the weights, you should own exactly how they are used. You still own the ramifications. This model will write what the base model would refuse. Intended for research, red-team, creative work without a hall monitor — and a threat if the prompt is. A disclaimer plus the payload is not a refusal. Over-abliteration into gibberish defeats the reason you wanted the unlock.

User Responsibility & Arbitration Clause

By accessing, downloading, using, running inference on, fine-tuning, merging, quantizing, distributing, integrating, or otherwise interacting with this model, you acknowledge and agree to the following:

1. Sole Responsibility. You are solely responsible for every prompt, every response, every downstream action, and any harm that results.
2. No Warranty. Provided "AS IS", without warranty of any kind.
3. Legal Compliance. You must comply with all applicable laws and policies in every jurisdiction you touch.
4. Operational Safety Layer. Deploy with input validation, output filtering, access controls, and human review for high-risk workflows. A production deploy without those layers is unsafe by construction.
5. Heightened Duty of Care. Uncensored means the duty of care sits with you. If unsure, do not make the request.
6. No Endorsement of Outputs. Authors do not endorse any specific output.
7. Arbitration. Disputes arising from use of this model are resolved by binding individual arbitration (AAA Consumer Rules if no other body is agreed), waiving jury trial and class actions.
8. Indemnification. You indemnify authors, contributors, and publishers against claims arising from your use.
9. Severability. Invalid provisions are replaced by the closest enforceable equivalent.
10. Acceptance. Use constitutes acceptance. If you do not accept, do not use the model.

This model is a tool with no opinions of its own. You supply the judgement. The outputs carry your fingerprints.

Provenance & Credits

• BF16 master: AEON-7/Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-BF16
• Base: Qwen/Qwen3.8-27B
• Quant: NVIDIA ModelOpt 0.46 mixed NVFP4+FP8 — AEON-7
• Serve: aeon-vllm-ultimate · aeon-vllm-ultimate-rtx

License

Apache 2.0 (inherited from Qwen/Qwen3.8-27B).