llmrun
AkahsizrrQwen 3.8Qwen3_5ForCausalLM

Qwen3.8 27B Code Tools Merged — Hardware Requirements & GPU Compatibility

ChatCodeReasoning

Qwen3.8 27B Code Tools Merged is a 26.9B-parameter open language model from Akahsizrr in the Qwen 3.8 family. It supports a context window of up to 262,144 tokens. At BF16 it needs about 54.54 GB of VRAM — see which GPUs and Macs can run it below.

26.9B params · dense262K contextApache 2.0331 downloads3 likes
Based on Qwen3.8 27B

Qwen3.8 27B Code Tools Merged runs on a 64 GB+ Mac or a workstation GPU

Best local pick → Mac Pro M2 Ultra (64 GB) · BF16 · 55 GB · ~10 tok/s

Macs from64 GB ~7 tok/s

Apache 2.0

Not set up for it?

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Qwen3.8 27B Code Tools Merged

One recommended way. Every other method, OS and quant is in the step-by-step guide.

Recommended

vLLM
one command · needs vLLM installed
$vllm serve Akahsizrr/Qwen3.8-27B-Code-Tools-Merged --max-model-len 16384

vLLM repo with its own quantisation

Download: Weights on Hugging Face

install, OS, other quants

Other ways

each opens step-by-step

Run in cloud

Fits on A100 80GB (25 GB headroom) · BF16

Generation speed
~24 tok/s
generation speed
Cost per 1M output tokens
$13.08
per 1M output tokens
Deployon Vast.aiBest optionfrom $1.14/hr (opens in new tab)
Compare GPUs →
or

How Much VRAM Does Qwen3.8 27B Code Tools Merged Need?

Select a quantization level; the listed hardware adheres to it.

QuantizationVRAM+32KOn RTX 4090 (reference)
54.5 GB55.9 GB16-bit · bfloat — Brain floating point 16 — preferred for trainingDoes not fit: needs 55 GB
est.calculated, no published file yet

Selected BF16

Brain floating point 16

55.0GBat 16K context

Weights
53.8 GB
KV cache
0.9 GB
Overhead
0.3 GB

Doesn't fit RTX 4090 (reference) · needs 31.0 GB more

≥10 tok/s needs ≥835 GB/s · M3 Ultra ~11, M2 Ultra ~10

Or rent A100 80GB from $1.14/hr.

Which Hardware Can Run Qwen3.8 27B Code Tools Merged?

BF16 · 54.5 GB

Qwen3.8 27B Code Tools Merged (BF16) requires 54.5 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 71+ GB is recommended. Using the full 262K context window can add up to 13.9 GB, bringing total usage to 68.4 GB. No consumer GPU has enough memory. 54 devices with unified memory can run Qwen3.8 27B Code Tools Merged, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, Mac Studio (M3 Ultra, 96GB).

By VRAM tier · typical GPUs and Macs at BF16
8–16 GB · RTX 4060, RTX 3060 12GB, RTX 5060 Ti 16GBDoesn't fit — smallest quant needs 55 GB8 GB tier →12 GB tier →16 GB tier →
24 GB · RTX 4090referenceDoesn't fit — smallest quant needs 55 GB24 GB tier →
Mac 32 GB · Mac Mini M4Doesn't fit — smallest quant needs 55 GBMacs →
Mac 64 GB · Mac Studio M4 MaxBF16 · 55 GB · ~7 tok/s · 64KMacs →

Alternatives to Qwen3.8 27B Code Tools Merged

Qwen 3.8 family · 12 models →

Similar models

6
ModelScoreVRAMtok/sWhy it is similar
Qwen3.6 27BAlibaba111.417.9 GB~38same size
Qwen3.5 27BAlibaba103.817.9 GB~38same size
Gemma 3 27B ITGoogle79.321.7 GB~36same size
Diffusiongemma 26B A4B ITGoogle—16.5 GB~171same size
Gemma 4 26B A4BGoogle—16.9 GB~153same size
Medgemma 27B ITGoogle—22.7 GB~34same size

Specifications

Publisher
Akahsizrr
Family
Qwen 3.8
Parameters
26.9B
Architecture
Qwen3_5ForCausalLM
Context Length
262,144 tokens
Vocabulary Size
248,320
Release Date
2026-09-10
License
Apache 2.0

Resources

Frequently Asked Questions

How much VRAM does Qwen3.8 27B Code Tools Merged need?

Qwen3.8 27B Code Tools Merged requires 54.5 GB of VRAM at BF16. Full 262K context adds up to 13.9 GB (68.4 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 26.9B × 16 bits ÷ 8 = 53.8 GB

KV Cache + Overhead ≈ 0.7 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 14.6 GB (at full 262K context)

VRAM usage by quantization

54.5 GB
68.4 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 5090 run Qwen3.8 27B Code Tools Merged?

No — Qwen3.8 27B Code Tools Merged requires at least 54.5 GB at BF16, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.

Can I run Qwen3.8 27B Code Tools Merged on a Mac?

Yes — Mac Mini M4 Pro (64 GB) and 40 other Macs can run Qwen3.8 27B Code Tools Merged. Apple Silicon uses unified memory, so the model shares RAM with the system. At BF16 you need at least 54.5 GB of usable unified memory (RAM minus macOS overhead).

Can I run Qwen3.8 27B Code Tools Merged locally?

Yes — Qwen3.8 27B Code Tools Merged can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Qwen3.8 27B Code Tools Merged?

At BF16, Qwen3.8 27B Code Tools Merged can reach ~88 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = (bandwidth GB/s ÷ model GB) × efficiency

Example: NVIDIA B200 → 8000 ÷ 54.5 × 0.65 = ~95 tok/s

Estimated speed at BF16 (54.5 GB)

~95 tok/s
~95 tok/s
~88 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of Qwen3.8 27B Code Tools Merged?

At BF16, the download is about 53.79 GB.

Which GPUs can run Qwen3.8 27B Code Tools Merged?

No single consumer GPU has enough VRAM to run Qwen3.8 27B Code Tools Merged at BF16 (54.5 GB). Multi-GPU or professional hardware is required.

Which devices can run Qwen3.8 27B Code Tools Merged?

54 devices with unified memory can run Qwen3.8 27B Code Tools Merged at BF16 (54.5 GB), including ASUS Ascent GX10, Asus ROG Flow Z13 (2025, Ryzen AI Max+ 395, 128 GB), Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.