Apertus 70B Instruct 2509 — Hardware Requirements & GPU Compatibility
Chat
4.9K downloads
182 likesSpecifications
- Publisher
- swiss-ai
- Parameters
- 70B
- Architecture
- ApertusForCausalLM
- Context Length
- 65,536 tokens
- Vocabulary Size
- 131,072
- Release Date
- 2025-11-14
- License
- Apache 2.0
Get Started
HuggingFace
How Much VRAM Does Apertus 70B Instruct 2509 Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| BF16 | 16.00 | 141.0 GB | 161.8 GB | 140.00 GB | Brain floating point 16 — preferred for training |
Which GPUs Can Run Apertus 70B Instruct 2509?
BF16 · 141.0 GBApertus 70B Instruct 2509 (BF16) requires 141.0 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 184+ GB is recommended. Using the full 66K context window can add up to 20.8 GB, bringing total usage to 161.8 GB. No single GPU has enough memory — multi-GPU or cluster setups are needed.
Which Devices Can Run Apertus 70B Instruct 2509?
BF16 · 141.0 GB4 devices with unified memory can run Apertus 70B Instruct 2509, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, Mac Pro M2 Ultra (192 GB).
Runs great
— Plenty of headroom
Decent
— Enough memory, may be tightRelated Models
Frequently Asked Questions
- How much VRAM does Apertus 70B Instruct 2509 need?
Apertus 70B Instruct 2509 requires 141.0 GB of VRAM at BF16. Full 66K context adds up to 20.8 GB (161.8 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 70B × 16 bits ÷ 8 = 140 GB
KV Cache + Overhead ≈ 1 GB (at 2K context + ~0.3 GB framework)
KV Cache + Overhead ≈ 21.8 GB (at full 66K context)
VRAM usage by quantization
BF16141.0 GBBF16 + full context161.8 GB- Can NVIDIA GeForce RTX 5090 run Apertus 70B Instruct 2509?
No — Apertus 70B Instruct 2509 requires at least 141.0 GB at BF16, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
- Can I run Apertus 70B Instruct 2509 on a Mac?
Apertus 70B Instruct 2509 requires at least 141.0 GB at BF16, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run Apertus 70B Instruct 2509 locally?
Yes — Apertus 70B Instruct 2509 can run locally on consumer hardware. At BF16 quantization it needs 141.0 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Apertus 70B Instruct 2509?
At BF16, Apertus 70B Instruct 2509 can reach ~21 tok/s on AMD Instinct MI300X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: AMD Instinct MI300X → 5300 ÷ 141.0 × 0.55 = ~21 tok/s
Estimated speed at BF16 (141.0 GB)
AMD Instinct MI300X~21 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Apertus 70B Instruct 2509?
At BF16, the download is about 140.00 GB.
- Which GPUs can run Apertus 70B Instruct 2509?
No single consumer GPU has enough VRAM to run Apertus 70B Instruct 2509 at BF16 (141.0 GB). Multi-GPU or professional hardware is required.
- Which devices can run Apertus 70B Instruct 2509?
4 devices with unified memory can run Apertus 70B Instruct 2509 at BF16 (141.0 GB), including Mac Pro M2 Ultra (192 GB), Mac Studio M2 Ultra (192 GB), NVIDIA DGX A100 640GB, NVIDIA DGX H100. Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.