Will It Run AI

AlibabaAlibaba

Qwen 3.6 27B

Frontier
huggingfaceHuggingFace
5.0MDownloads1.5KLikesApr 2026Veröffentlicht262K TokenKontextApache 2.0Lizenz99 HerausragendQualität

Qwen 3.6 27B (27B parameters) requires approximately 18.9 GB of VRAM with Q4_K_M quantization. For the best balance of quality and speed, we recommend hardware with at least 22 GB of VRAM.

Loslegen

— kopieren & einfügen, um lokal auszuführen

Copy-paste commands to run Qwen 3.6 27B on your machine.

Run

lms load Qwen3.6-27B && lms server start

Quick specs

Parameters27B
Architecturedense
Context262K tokens
Modalitytext+vision
Min RAM10.5 GB
Rec. RAM16.5 GB (Q4_K_M)
LicenseApache 2.0
FamilyQwen
✓ Vision✓ Tool Use✓ Thinking✓ Code✓ Chat✓ Reasoning

About this model

Qwen 3.6 27B is a dense, multimodal 27B model that beats the previous-gen Qwen 3.5 397B-A17B MoE flagship on SWE-bench Verified (77.2% vs 76.2%) while fitting on a 16-24 GB consumer GPU. It uses a Gated DeltaNet + Gated Attention hybrid, ships a vision encoder, and supports 262K native context extensible to ~1M tokens via YaRN.

  • •Dense 27B beats Qwen 3.5 397B-A17B MoE on SWE-bench Verified (77.2%)
  • •Fits on RTX 4080/4090/3090 (16-24 GB) at Q4_K_M (~16.8 GB)
  • •Multimodal: vision encoder for images, OCR, and hour-scale video
  • •262K native context, extensible to ~1M tokens via YaRN

Verwandte Modelle

Deine Hardware

Erkennung...

Schnellauswahl

Intel
Bestes BudgetS
Intel Arc Pro B60 24GB~$599 — 12 tok/s
NVIDIA
Beste GesamtwahlS
RTX 5090 32GB~$1,999 — 79 tok/s

Beste Hardware

Top-Empfehlungen für Qwen 3.6 27B

NVIDIA
RTX 5090 32GBS
32 GB
NVIDIA
NVIDIA A100 40GBS
40 GB
NVIDIA
RTX PRO 4500 Blackwell 32GBS
32 GB
AMD Instinct MI100 32GBS
32 GB
MacBook Pro M4 Max 48GBS
48 GB

Dieses Modell ausführen

Qwen 3.6 27B on RTX 5090 32GBQwen 3.6 27B on NVIDIA A100 40GBQwen 3.6 27B on RTX PRO 4500 Blackwell 32GB

Quantisierungsoptionen

VRAM-Schätzungen nach Quantisierungsstufe

No hardware detected — fit column shows raw VRAM estimates

QuantBitsVRAMQualityFit
Q2_K
2
10.5 GB
Low—
Q3_K_S
3
13.2 GB
Low—
NVFP4
4
15.1 GB
Medium—
Q4_K_M
4
16.5 GB
Medium—
Q5_K_M
5
19.4 GB
High—
Q6_K
6
22.1 GB
High—
Q8_0
8
28.9 GB
Very High—
F16
16
55.4 GB
Maximum—

Quality benchmarks

Qwen 3.6 27B benchmark scores

Benchmark verified

Coding

SWE-bench Verified77.2%
HumanEval+—
Aider Polyglot—
LiveCodeBench83.9%

Reasoning

MMLU-Pro86.2%
GPQA Diamond87.8%
MATH-500—
ARC Challenge—

Source: official · 2026-04-22

Hardware-Kompatibilität

Eignungsschätzungen für alle Hardware

Rechner öffnen

Computing compatibility...

Speicheraufschlüsselung

Reference: RTX 2060 6GB

Weights16.5 GB
KV Cache1.0 GB
Runtime0.9 GB
Headroom0.6 GB

Häufig gestellte Fragen

FAQ — Qwen 3.6 27B

Siehe auch

QuantisierungsleitfadenBewertungsmethodikRechner öffnen