Mellum 4B Base — Hardware Requirements & GPU Compatibility
ChatCode
Mellum 4B Base is a 4.0B-parameter open language model from JetBrains. It supports a context window of up to 8,192 tokens. At BF16 it needs about 9.09 GB of VRAM — see which GPUs and Macs can run it below.
3.2K downloads
447 likesSpecifications
- Publisher
- JetBrains
- Parameters
- 4.0B
- Architecture
- LlamaForCausalLM
- Context Length
- 8,192 tokens
- Vocabulary Size
- 98,304
- Release Date
- 2025-05-07
- License
- Apache 2.0
Get Started
HuggingFace
How Much VRAM Does Mellum 4B Base Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| BF16 | 16.00 | 9.1 GB | 11.4 GB | 8.04 GB | Brain floating point 16 — preferred for training |
Which GPUs Can Run Mellum 4B Base?
BF16 · 9.1 GBMellum 4B Base (BF16) requires 9.1 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 12+ GB is recommended. Using the full 8K context window can add up to 2.3 GB, bringing total usage to 11.4 GB. 28 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti, NVIDIA GeForce RTX 3080 Ti.
Runs great
— Plenty of headroom
Decent
— Enough VRAM, may be tightWhich Devices Can Run Mellum 4B Base?
BF16 · 9.1 GB27 devices with unified memory can run Mellum 4B Base, including NVIDIA DGX H100, NVIDIA DGX A100 640GB.
Runs great
— Plenty of headroomRelated Models
Frequently Asked Questions
- How much VRAM does Mellum 4B Base need?
Mellum 4B Base requires 9.1 GB of VRAM at BF16. Full 8K context adds up to 2.3 GB (11.4 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 4.0B × 16 bits ÷ 8 = 8 GB
KV Cache + Overhead ≈ 1.1 GB (at 2K context + ~0.3 GB framework)
KV Cache + Overhead ≈ 3.4 GB (at full 8K context)
VRAM usage by quantization
BF169.1 GBBF16 + full context11.4 GB- Can I run Mellum 4B Base on a Mac?
Mellum 4B Base requires at least 9.1 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 Mellum 4B Base locally?
Yes — Mellum 4B Base can run locally on consumer hardware. At BF16 quantization it needs 9.1 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Mellum 4B Base?
At BF16, Mellum 4B Base can reach ~321 tok/s on AMD Instinct MI300X. On NVIDIA GeForce RTX 4090: ~72 tok/s. 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 ÷ 9.1 × 0.55 = ~321 tok/s
Estimated speed at BF16 (9.1 GB)
~321 tok/s~72 tok/s~240 tok/s~198 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Mellum 4B Base?
At BF16, the download is about 8.04 GB.
- Which GPUs can run Mellum 4B Base?
28 consumer GPUs can run Mellum 4B Base at BF16 (9.1 GB). Top options include AMD Radeon RX 6800, AMD Radeon RX 6800 XT, AMD Radeon RX 6900 XT, AMD Radeon RX 6700 XT. 17 GPUs have plenty of headroom for comfortable inference.
- Which devices can run Mellum 4B Base?
27 devices with unified memory can run Mellum 4B Base at BF16 (9.1 GB), including Mac Mini M4 (16 GB), Mac Mini M4 (32 GB), Mac Mini M4 Pro (24 GB), Mac Mini M4 Pro (48 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.