EuroMoE 2.6B A0.6B Instruct 2512 — Hardware Requirements & GPU Compatibility
Chat
EuroMoE 2.6B A0.6B Instruct 2512 is a 2.6B-parameter open language model from utter-project. It supports a context window of up to 32,768 tokens. At Q4_K_M it needs about 1.92 GB of VRAM — see which GPUs and Macs can run it below.
Runs great on a 24 GB GPU
Runs from 8 GB (Q4_K_M · 32K)Best on 24 GB (Q4_K_M · ~532 tok/s · 32K)Macs from 8 GB (Q4_K_M · ~101 tok/s)
Apache 2.0
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run EuroMoE 2.6B A0.6B Instruct 2512
One recommended way. Every other method, OS and quant is in the step-by-step guide.
Recommended
vllm serve utter-project/EuroMoE-2.6B-A0.6B-Instruct-2512 --max-model-len 16384vLLM repo with its own quantisation
Download: Weights on Hugging Face
Other ways
each opens step-by-stepRun in cloud
Fits on RTX 3060 12GB (9 GB headroom) · Q4_K_M
- Generation speed
- ~347 tok/s
- generation speed
- Cost per 1M output tokens
- $0.061
- per 1M output tokens
How Much VRAM Does EuroMoE 2.6B A0.6B Instruct 2512 Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 1.5 GB | 2.2 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 32K context · ~573 tok/s | |
| 1.6 GB | 2.4 GB | 3-bit medium — 3-bit medium quantization | Fits: 32K context · ~558 tok/s | |
| 1.9 GB | 2.7 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 32K context · ~532 tok/s | |
| 2.2 GB | 3 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Fits: 32K context · ~510 tok/s | |
| 2.5 GB | 3.3 GB | 6-bit, very good — 6-bit quantization, very good quality | Fits: 32K context · ~489 tok/s | |
| 3.0 GB | 3.7 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Fits: 32K context · ~459 tok/s | |
| 5.6 GB | 6.3 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Fits: 32K context · ~340 tok/s |
Selected Q4_K_M
4-bit medium quantization2.3GBat 16K context
- Weights
- 1.6 GB
- KV cache
- 0.4 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · 21.7 GB free · ~532 tok/s
≥50 tok/s needs ≥30 GB/s · RTX 5090 ~612, RTX 3090 Ti ~532
Which Hardware Can Run EuroMoE 2.6B A0.6B Instruct 2512?
Q4_K_M · 1.9 GBEuroMoE 2.6B A0.6B Instruct 2512 (Q4_K_M) requires 1.9 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 3+ GB is recommended. Using the full 33K context window can add up to 0.8 GB, bringing total usage to 2.7 GB. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 163 devices with unified memory can run EuroMoE 2.6B A0.6B Instruct 2512, including NVIDIA DGX Spark, NVIDIA Jetson AGX Thor Developer Kit.
≤ 11 GB
≤ 11 GB tier →fits · 5.7 GB free · up to 32K ctx
16 GPUs · 1 Mac class · 2 device families
RTX 3080~485Simulate output on RTX 3080RTX 2080 Ti~448Simulate output on RTX 2080 TiRTX 3070 Ti~445Simulate output on RTX 3070 TiM3*8 GB~101Simulate output on iMac 24" M3 (8 GB)Orin Nano~146Simulate output on NVIDIA Jetson Orin Nano 8GB (Super)iPhone~80Simulate output on iPhone 1720 GB
fits · 17.7 GB free · up to 32K ctx
1 GPU
RX 7900 XT~180Simulate output on RX 7900 XT32–40 GB
fits · 29.7 GB free · up to 32K ctx
1 GPU · 3 Mac classes · 4 device families
RTX 5090~612Simulate output on RTX 5090M5 Max36 GB~169Simulate output on Mac Studio M5 Max (36 GB)M532 GB~122Simulate output on MacBook Air 13" M5 (32 GB)Orin~246Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~152Simulate output on Snapdragon X Elite Copilot+ PC
By VRAM tier · typical GPUs and Macs at Q4_K_M
Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| MiniCPM5 2BOpenBMB | — | 2.5 GB | ~345 | same size |
| North Micro Vision InstructCohere | — | 7.1 GB | ~119 | same size |
| LFM2.5 2.6BLiquid AI | — | 2.3 GB | ~321 | same size |
| Granite 3.1 2B InstructIBM | — | 3.2 GB | ~329 | same size |
| Gemma 2 2B ITGoogle | — | 2.5 GB | ~379 | same size |
| Phi 2Microsoft | — | 2.6 GB | ~248 | same size |
Specifications
- Publisher
- utter-project
- Parameters
- 2.6B
- Active parameters
- 0.5B per token
- Architecture
- MixtralForCausalLM
- Context Length
- 32,768 tokens
- Vocabulary Size
- 128,000
- Release Date
- 2025-12-15
- License
- Apache 2.0
Frequently Asked Questions
How much VRAM does EuroMoE 2.6B A0.6B Instruct 2512 need?
EuroMoE 2.6B A0.6B Instruct 2512 requires 1.9 GB of VRAM at Q4_K_M, or 5.6 GB at BF16. Full 33K context adds up to 0.8 GB (2.7 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 2.6B × 4.8 bits ÷ 8 = 1.6 GB
KV Cache + Overhead ≈ 0.3 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 ≈ 1.1 GB (at full 33K context)
VRAM usage by quantization
What's the best quantization for EuroMoE 2.6B A0.6B Instruct 2512?
For EuroMoE 2.6B A0.6B Instruct 2512, Q4_K_M (1.9 GB) offers the best balance of quality and VRAM usage. Q5_K_M (2.2 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 1.5 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run EuroMoE 2.6B A0.6B Instruct 2512 on a Mac?
Yes — MacBook Air 13" M3 (8 GB) and 145 other Macs can run EuroMoE 2.6B A0.6B Instruct 2512. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 1.9 GB of usable unified memory (RAM minus macOS overhead).
Can I run EuroMoE 2.6B A0.6B Instruct 2512 locally?
Yes — EuroMoE 2.6B A0.6B Instruct 2512 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is EuroMoE 2.6B A0.6B Instruct 2512?
At Q4_K_M, EuroMoE 2.6B A0.6B Instruct 2512 can reach ~205 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~532 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = 1000 ÷ (active GB ÷ (bandwidth GB/s × efficiency) × 1000 + layers × routing ms)
Mixture-of-Experts: just the active experts get read for each token, along with a fixed routing cost per layer.
Example: NVIDIA B200 → 0.4 GB active ÷ (8000 × 0.65) = 0.07 ms, plus 24 layers × 0.055 ms = 1.32 ms, so 1000 ÷ 1.39 ms = ~719 tok/s
Estimated speed at Q4_K_M (1.9 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of EuroMoE 2.6B A0.6B Instruct 2512?
At Q4_K_M, the download is about 1.57 GB. The full-precision BF16 version is 5.22 GB. The smallest option (Q2_K) is 1.11 GB.
Which GPUs can run EuroMoE 2.6B A0.6B Instruct 2512?
52 consumer GPUs can run EuroMoE 2.6B A0.6B Instruct 2512 at Q4_K_M. Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT. 52 GPUs have plenty of headroom for comfortable inference.
Which devices can run EuroMoE 2.6B A0.6B Instruct 2512?
163 devices with unified memory can run EuroMoE 2.6B A0.6B Instruct 2512 at Q4_K_M (1.9 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, 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.