Qwen3.6-27B-AEON-Ultimate-Uncensored-Multimodal-NVFP4-MTP

Deployment, operations & benchmarks → github.com/AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-DFlash

The GitHub repo is the source of truth for the production deployment guide, hardware-tuned docker-compose configs, full configuration reference, measured benchmarks, and AGENTS.md — an operator's manual that pre-empts common stale-documentation traps.

🙏 Reference recipe credit: The modelopt + MTP graft pipeline used to build this variant is based on sakamakismile's validated Qwen3.6-27B-NVFP4-MTP series (22K+ downloads). They worked out the modelopt config, the per-projection quantization choices, and the MTP-head graft technique on the un-abliterated base; we adapted the same recipe to AEON-Ultimate's abliterated weights. The reference benchmark numbers cited below are theirs. Full credit for the recipe → sakamakismile.

Variants

Format	Size	Use case
BF16	51 GB	Full-precision reference weights (A100/H100 80 GB, RTX PRO 6000 96 GB, multi-GPU, fine-tuning)
NVFP4 (compressed-tensors + DFlash)	26 GB	DGX Spark / GB10 — production validated with DFlash speculative decoding. Patched vllm-aeon-ultimate-dflash container.
Multimodal-NVFP4-MTP (this repo)	27 GB	High-bandwidth dedicated GPUs (RTX 5090, RTX PRO 6000, B100/B200) with MTP speculative decoding via the model's native mtp.* head. modelopt format, --quantization modelopt. Vision tower preserved.
Text-NVFP4-MTP	20 GB	Same as this repo but with vision tower stripped. Smaller footprint for text-only deployments on tighter VRAM.

What this is

This is the modelopt-format NVFP4 variant with MTP speculative decoding, multimodal-preserved, of AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-BF16 — the lossless abliteration of Qwen 3.6 27B (KL 0.000492 vs base, 0/100 refusals, multimodal preserved, hybrid GDN-aware quantization).

Specifically:

• Body quantized to NVFP4 via nvidia-modelopt 0.43.0 with NVFP4_DEFAULT_CFG. This is the modelopt compressed-tensors format that vLLM serves through --quantization modelopt (different code path from the -NVFP4 sibling release which uses --quantization compressed-tensors).
• Linear-attn / GatedDeltaNet layers preserved BF16 (432 keys across 48 GDN layers). NVFP4 quantization on Mamba/SSM state collapses the recurrence; modelopt's *linear_attn.conv1d* ignore plus our explicit *linear_attn* exclude keeps these intact.
• Vision tower preserved BF16 (333 keys). Multimodal inference fully functional.
• MTP head grafted from the base Qwen/Qwen3.6-27B checkpoint (15 tensors, BF16). The base contains MTP heads but Qwen3_5ForConditionalGeneration.from_pretrained drops them during loading; the lna-lab pipeline pattern (which this build follows) explicitly grafts them back into the quantized output, giving vLLM a working drafter for --speculative-config '{"method":"qwen3_5_mtp",...}'.

Why MTP — and where it actually wins

Multi-Token Prediction (MTP) lets the model predict multiple future tokens per forward pass via the trained mtp.* head, enabling speculative decoding without a separate drafter model. The acceptance rate is high because the drafter is the model itself — same architecture, same weights, same distribution.

Measured numbers on AEON-Ultimate (this exact variant)

Hardware	Median tok/s	Peak tok/s	Spec-decode acceptance
RTX PRO 6000 Blackwell (96 GB dedicated VRAM)	~92 (this variant) / 111.4 (XS sibling)	124.7 (XS sibling)	67.7 % regular / 69.2 % XS
DGX Spark / GB10 (unified memory)	24.1 (XS sibling)	27.5	66.3 %
RTX 5090, B100 / B200	not yet measured by us — community welcome

Reference numbers from sakamakismile's un-abliterated recipe (RTX 5090)

• Single-stream short prompts at n=3: ~132 tok/s
• Single-stream long-form: ~105 tok/s
• 2-parallel aggregate (256K + KV FP8): ~189–207 tok/s
• Mean MTP acceptance length: ~3.0–4.0 (vs DFlash chains ~2.0–2.3)

The hardware-routing punchline

On RTX PRO 6000 the XS sibling beats DFlash territory (~111 tok/s vs DFlash-class ~85 we'd expect there). On DGX Spark, DFlash beats MTP by 26 % median / 52 % peak — the unified-memory bandwidth caps how much MTP's high acceptance can translate to throughput. So: MTP is a dedicated-VRAM-Blackwell variant, not a universal upgrade. Full bench data: GitHub repo Performance section.

🎯 When to pick this variant — measured hardware routing

The right speculative-decode method depends on memory architecture:

Hardware tier	Recommended variant	Why
DGX Spark / GB10 (sm_121a, unified memory)	-NVFP4 (DFlash) — not this MTP variant	Bench on Spark: DFlash beats MTP by +26 % median, +52 % peak. Spark's unified-memory bandwidth doesn't reward MTP's high acceptance rate. Don't run MTP on Spark.
RTX PRO 6000 Blackwell (sm_120, 96 GB dedicated VRAM)	This variant (Multimodal-NVFP4-MTP) ✅ if you need vision; Text if text-only	MTP wins on dedicated VRAM. ~92 tok/s median measured with GDN BF16; dedicated-VRAM bandwidth lets the MTP head's high acceptance rate translate to throughput.
RTX 5090 (sm_120, 32 GB dedicated VRAM)	Multimodal-XS if you use vision; Text-XS if text-only	XS variants fit comfortably in 32 GB. 111.4 tok/s median measured on RTX PRO 6000; RTX 5090 should land near or above that.
A100 / H100 (no native FP4)	BF16	NVFP4 dequantizes to BF16 on Ampere/Hopper — no benefit.
B100 / B200 (sm_100, dedicated FP4)	This variant (Multimodal) or Text variant	Native FP4 + dedicated VRAM = MTP territory.

Full bench numbers: GitHub repo Performance section.

Usage

vLLM serve

# One-time: pull this repo locally
hf download AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-Multimodal-NVFP4-MTP \
  --local-dir ./aeon-ultimate-multimodal-nvfp4-mtp

# Serve
export VLLM_NVFP4_GEMM_BACKEND=flashinfer-cutlass
export VLLM_USE_FLASHINFER_MOE_FP4=0
export VLLM_USE_FLASHINFER_SAMPLER=1

vllm serve ./aeon-ultimate-multimodal-nvfp4-mtp \
  --quantization modelopt \
  --trust-remote-code \
  --max-model-len 262144 \
  --max-num-seqs 32 \
  --max-num-batched-tokens 32768 \
  --gpu-memory-utilization 0.94 \
  --enable-chunked-prefill \
  --enable-prefix-caching \
  --reasoning-parser qwen3 \
  --tool-call-parser qwen3_coder \
  --enable-auto-tool-choice \
  --speculative-config '{"method":"qwen3_5_mtp","num_speculative_tokens":3}'

num_speculative_tokens=3 is the canonical setting for qwen3_5_mtp. Higher values diverge the drafter further from the target distribution and acceptance falls.

Configuration notes

• --quantization modelopt is required (not compressed-tensors — different format).
• --speculative-config '{"method":"qwen3_5_mtp", ...}' activates the grafted MTP head as the spec-decode drafter. No external drafter download needed — the head is in the safetensors of this repo.
• --gpu-memory-utilization 0.94 is the validated cap on RTX PRO 6000; 0.95 causes the FlashInfer NVFP4 GEMM autotuner to OOM on first boot. See the GitHub repo's RTX PRO 6000 page for the same OOM behavior under DFlash.

Quantization recipe

• Tool: nvidia-modelopt 0.43.0 with NVFP4_DEFAULT_CFG
• Loader: Qwen3_5ForConditionalGeneration.from_pretrained (multimodal-preserved class)
• Calibration: neuralmagic/calibration LLM split, 20 samples × 8192 tokens
• Excluded from quantization (kept BF16):
  • lm_head, proj_out.*, *router*, *mlp.gate.* (NVFP4_DEFAULT_CFG)
  • *linear_attn.conv1d*, *mixer.conv1d* (NVFP4_DEFAULT_CFG)
  • *linear_attn* (added — full GDN preservation)
  • *visual* (added — vision tower preservation)
  • *mtp* (added — MTP head preservation)
  • *output_layer*, output.*
• MTP graft: 15 tensors copied bf16 from Qwen/Qwen3.6-27B after modelopt export (AutoModelForCausalLM.from_pretrained drops them; explicit graft restores)
• Pipeline: lna-lab/GGUF-to-NVFP4-SM120 reference recipe, adapted for AEON-Ultimate-BF16 input + separate MTP source

Provenance & credits

• BF16 source: AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-BF16. See that card for the full abliteration pipeline.
• MTP graft technique: lna-lab/GGUF-to-NVFP4-SM120 (docs/MTP_GRAFT_RECIPE.md)
• Reference benchmark recipes: sakamakismile/Qwen3.6-27B-Text-NVFP4-MTP
• Quantization: NVIDIA TensorRT Model Optimizer (nvidia-modelopt 0.43.0)
• Base: Alibaba Qwen team — Qwen/Qwen3.6-27B

License + responsibility

Apache 2.0, inherited from Qwen/Qwen3.6-27B. This is an uncensored model. Read the full User Responsibility & Arbitration Clause on the BF16 source card before deploying. Summary: you implement downstream safety layers (input validation, output filtering, content moderation, audit logging, rate limiting, access controls, human-in-the-loop for high-risk workflows). The model has no opinions of its own — you supply the opinions, the judgment, and the ethics.