EuroLLM 22B — VRAM, speed & local setup
Written by Jakub Rusinowski · Last updated
The flagship EuroLLM, reported to match or beat global models of similar size on multilingual tasks. Fits a 16 GB card at Q4.
EuroLLM 22B needs about 14 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache. On Apple Silicon that figure comes out of unified memory.
VRAM and speed by quantization
Quoted against NVIDIA RTX 4090 (24 GB). Weights plus overhead only: this model's architecture is unpublished, so no KV cache is added.
| Quant | Memory | VRAM | Speed (est.) | Fit |
|---|---|---|---|---|
| Q2_K 2.63 bpw | 8 GB | ~78 tok/s | Fits | |
| Q3_K_M 3.41 bpw | 10.2 GB | ~64 tok/s | Fits | |
| Q4_K_M 4.83 bpw | 14.1 GB | ~48 tok/s | Fits | |
| Q5_K_M 5.67 bpw | 16.4 GB | ~42 tok/s | Fits | |
| Q6_K 6.56 bpw | 18.8 GB | ~37 tok/s | Fits | |
| Q8_0 8.50 bpw | 24.2 GB | ~4 tok/s | Offload | |
| F16 16.00 bpw | 44.8 GB | ~2 tok/s | Offload |
Black marker = usable memory on the NVIDIA RTX 4090 (24 GB). Estimates from the memory-bandwidth roofline on the methodology page. EuroLLM 22B VRAM calculator →
Get EuroLLM 22B running
The cheapest catalogued GPU that runs EuroLLM 22B is the AMD Radeon RX 9060 XT 16GB (16 GB).
or compare on Vast.ai from $0.35/hr (typical low · varies)
As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.
How to run EuroLLM 22B
Install Ollama, then run:
ollama run eurollmSpecifications
Corroborated — Two or more independent sources agree on these figures, but the model card itself was not retrieved. Treat the numbers as good rather than confirmed. Still unconfirmed: context.
- Parameters
- 22 Billion
- Context window
- 4,096
- Architecture
- Dense
- Provider
- Unbabel / Instituto Superior Técnico / INESC-ID
- Licence
- Apache 2.0
- Practical deployment
- Laptop
- Specified at
- Q4_K_M
- System RAM
- 32 GB
- Record updated
- 2026-09-19
Commercial use permitted. No usage restrictions beyond attribution.
Quality and use cases
Scores as published by the model’s authors or an independent evaluator — quality, not throughput, and not measured by us.
Other EuroLLM sizes
EuroLLM 22B — frequently asked questions
How much VRAM does EuroLLM 22B need?
About 14 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.
Does EuroLLM 22B run on an RTX 4090 (24 GB)?
Yes. EuroLLM 22B needs about 14 GB at Q4_K_M, inside a 24 GB card, at an estimated 48 tokens/sec.
How do I run EuroLLM 22B locally?
Install Ollama and run `ollama run eurollm`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.
What other sizes does EuroLLM come in?
EuroLLM 9B (6 GB), EuroLLM 22B (14 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.