llmrun
swiss-aiApertusApertusForCausalLM

Apertus 8B 2509 — Hardware Requirements & GPU Compatibility

Chat

Apertus 8B 2509 is a 8.1B-parameter open language model from swiss-ai in the Apertus family. It supports a context window of up to 65,536 tokens. At Q4_K_M it needs about 5.40 GB of VRAM — see which GPUs and Macs can run it below.

8.1B params · dense66K contextApache 2.042.7K downloads175 likes
S

Runs great on a 24 GB GPU

Runs from8 GB Q4_K_M · 16KBest on24 GB Q4_K_M · ~121 tok/s · 64KMacs from16 GB Q4_K_M · ~26 tok/s

Apache 2.0

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Apertus 8B 2509

One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.

Recommended

vLLM
one command · needs vLLM installed
$vllm serve swiss-ai/Apertus-8B-2509 --max-model-len 16384

vLLM repo with its own quantisation

Download: Weights on Hugging Face

install, OS, other quants

Ways to runOther ways

each opens step-by-step

Run in cloud

Runs on RTX 3080 (10 GB) (2 GB headroom) · Q4_K_M

Generation speed
~92 tok/s
generation speed
Cost per 1M output tokens
$0.37
per 1M output tokens

How Much VRAM Does Apertus 8B 2509 Need?

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

QuantizationVRAM+32KOn RTX 4090 (reference)
4.0 GB8 GB2-bit · K-quant — 2-bit quantization with K-quant improvementsFits: 64K context · ~164 tok/s
4.5 GB8.5 GB3-bit medium — 3-bit medium quantizationFits: 64K context · ~146 tok/s
5.4 GB9.4 GB4-bit medium — 4-bit medium quantization — most popular sweet spotFits: 64K context · ~121 tok/s
6.3 GB10.3 GB5-bit medium — 5-bit medium quantization — good quality/size tradeoffFits: 64K context · ~104 tok/s
7.2 GB11.2 GB6-bit, very good — 6-bit quantization, very good qualityFits: 64K context · ~91 tok/s
8.6 GB12.7 GB8-bit, near-lossless — 8-bit quantization, near-losslessFits: 64K context · ~76 tok/s
16.7 GB20.7 GB16-bit · bfloat — Brain floating point 16 — preferred for trainingFits: 32K context · ~39 tok/s
est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

7.3GBat 16K context

Weights
4.8 GB
KV cache
2.2 GB
Overhead
0.3 GB

Fits RTX 4090 (reference) · 16.7 GB free · ~121 tok/s

≥50 tok/s needs ≥415 GB/s · RTX 5090 ~216, M5 Ultra ~159

Which Hardware Can Run Apertus 8B 2509?

Q4_K_M · 5.4 GB

Apertus 8B 2509 (Q4_K_M) requires 5.4 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 8+ GB is recommended. Using the full 66K context window can add up to 8.3 GB, bringing total usage to 13.7 GB. 107 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti, NVIDIA GeForce RTX 3070 Ti. 172 devices with unified memory can run Apertus 8B 2509, including Mac Studio M5 Ultra (256 GB), Mac Studio M5 Ultra (512 GB), NVIDIA Jetson Orin Nano 8GB (Super).

By VRAM tier · typical GPUs and Macs at Q4_K_M
8 GB · RTX 4060Q4_K_M · 7.3 GB · ~33 tok/s · 16K8 GB tier →
12 GB · RTX 3060 12GBQ4_K_M · 7.3 GB · ~43 tok/s · 32K12 GB tier →
16 GB · RTX 5060 Ti 16GBQ4_K_M · 7.3 GB · ~54 tok/s · 64K16 GB tier →
24 GB · RTX 4090referenceQ4_K_M · 7.3 GB · ~121 tok/s · 64K24 GB tier →
Mac 32 GB · Mac Mini M4Q4_K_M · 7.3 GB · ~16 tok/s · 64KMacs →
Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 7.3 GB · ~71 tok/s · 64KMacs →

Alternatives to Apertus 8B 2509

Apertus family · 5 models →

Similar models

6
ModelScoreVRAMtok/sWhy it is similar
ALoDLM 8BAmazon—7.6 GB~119same size
Llm Jp 4.1 8B Thinkingllm-jp—7.6 GB~115same size
Ling 3.0 TinyInclusion AI—5.5 GB~364same size
LFM2.5 8B A1BLiquid AI—5.6 GB~335same size
BitCPM CANN 8BOpenBMB—5.6 GB~127same size
Gemma 4 E4BGoogle—5.6 GB~123same size
Fine-tunes & variants(1)

Specifications

Publisher
swiss-ai
Family
Apertus
Parameters
8.1B
Architecture
ApertusForCausalLM
Context Length
65,536 tokens
Vocabulary Size
131,072
Release Date
2025-09-02
License
Apache 2.0

Frequently Asked Questions

How much VRAM does Apertus 8B 2509 need?

Apertus 8B 2509 requires 5.4 GB of VRAM at Q4_K_M, or 16.7 GB at BF16. Full 66K context adds up to 8.3 GB (13.7 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 8.1B × 4.8 bits ÷ 8 = 4.8 GB

KV Cache + Overhead ≈ 0.6 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 ≈ 8.9 GB (at full 66K context)

VRAM usage by quantization

5.4 GB
13.7 GB

Learn more about VRAM estimation →

What's the best quantization for Apertus 8B 2509?

For Apertus 8B 2509, Q4_K_M (5.4 GB) offers the best balance of quality and VRAM usage. Q5_K_M (6.3 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 4.0 GB.

VRAM requirement by quantization

Q2_K
4.0 GB
Q4_K_M ★
5.4 GB
Q5_K_M
6.3 GB
Q6_K
7.2 GB
Q8_0
8.6 GB
BF16
16.7 GB

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Apertus 8B 2509 on a Mac?

Yes — Mac Mini M2 Pro (16 GB) and 145 other Macs can run Apertus 8B 2509. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 5.4 GB of usable unified memory (RAM minus macOS overhead).

Can I run Apertus 8B 2509 locally?

Yes — Apertus 8B 2509 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Apertus 8B 2509?

At Q4_K_M, Apertus 8B 2509 can reach ~889 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~121 tok/s. 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 ÷ 5.4 × 0.65 = ~963 tok/s

Estimated speed at Q4_K_M (5.4 GB)

~963 tok/s
~121 tok/s
~963 tok/s
~889 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 Apertus 8B 2509?

At Q4_K_M, the download is about 4.83 GB. The full-precision BF16 version is 16.11 GB. The smallest option (Q2_K) is 3.42 GB.

Which GPUs can run Apertus 8B 2509?

107 consumer GPUs can run Apertus 8B 2509 at Q4_K_M. Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6700M, AMD Radeon RX 6750 XT, AMD Radeon RX 5600 XT. 51 GPUs have plenty of headroom for comfortable inference.

Which devices can run Apertus 8B 2509?

172 devices with unified memory can run Apertus 8B 2509 at Q4_K_M (5.4 GB), including AMD Ryzen 7 8845HS (Hawk Point) Laptop, AMD Ryzen AI 7 350 (Krackan Point) Laptop, AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10. Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.