Moondream2 — Hardware Requirements & GPU Compatibility
Vision
Moondream2 is a compact 1.9-billion-parameter vision-language model from the Moondream project, published under developer Vikhyat Korrapati's vikhyatk account, designed for lightweight image understanding rather than open-ended chat. It handles image captioning, visual question answering, object detection, and point-based visual grounding, useful for document pipelines and robotics perception where a full-size multimodal model is unnecessary. Its small size means it runs comfortably on modest consumer GPUs and even CPU-only or edge setups. Because Moondream2 focuses on single-image and short-prompt tasks, it does not advertise an extended context window like larger chat models. It is released under the Apache 2.0 license, allowing unrestricted commercial and research use. First published in March 2024, it remains a popular choice for on-device and edge vision applications, with built-in object detection and pointing alongside captioning.
Runs great on a 24 GB GPU
Runs from 8 GB (Q4_K_M · 32K)Best on 24 GB (Q4_K_M · ~516 tok/s · 32K)Macs from 8 GB (Q4_K_M · ~56 tok/s)
Apache 2.0
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Moondream2
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
ollama run hf.co/ggml-org/moondream2-20250414-GGUFQ4_K_M fits fully on the RTX 4090 · 21.0 GB free · 16K context
Download: moondream2-20250414-Q4_K_M.gguf1.16 GBggml-org
Ways to runOther ways
each opens step-by-stepRun in cloud
Runs on RTX 3080 Ti (12 GB) (9 GB headroom) · Q4_K_M
- Generation speed
- ~467 tok/s
- generation speed
- Cost per 1M output tokens
- $0.073
- per 1M output tokens
How Much VRAM Does Moondream2 Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 0.9 GB | 4.6 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 32K context · ~728 tok/s | |
| 1.0 GB | 4.7 GB | 3-bit medium — 3-bit medium quantization | Fits: 32K context · ~636 tok/s | |
| 1.3 GB | 5 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 32K context · ~516 tok/s | |
| 1.5 GB | 5.2 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Fits: 32K context · ~434 tok/s | |
| 1.8 GB | 5.4 GB | 6-bit, very good — 6-bit quantization, very good quality | Fits: 32K context · ~374 tok/s | |
| 2.1 GB | 5.8 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Fits: 32K context · ~309 tok/s | |
| 4.2 GB | 7.9 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Fits: 32K context · ~155 tok/s |
Selected Q4_K_M
4-bit medium quantization3.0GBat 16K context
- Weights
- 0.7 GB
- KV cache
- 2.0 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · 21.0 GB free · ~516 tok/s
≥50 tok/s needs ≥98 GB/s · RTX 5090 ~917, RTX 3090 Ti ~516
Which Hardware Can Run Moondream2?
Q4_K_M · 1.3 GBMoondream2 (Q4_K_M) requires 1.3 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 2+ GB is recommended. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 163 devices with unified memory can run Moondream2, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB).
≤ 11 GB
≤ 11 GB tier →fits · 5.0 GB free · up to 32K ctx
16 GPUs · 1 Mac class · 2 device families
RTX 3080~389Simulate output on RTX 3080RTX 2080 Ti~315Simulate output on RTX 2080 TiRTX 3070 Ti~311Simulate output on RTX 3070 TiM3*8 GB~56Simulate output on iMac 24" M3 (8 GB)Orin Nano~52Simulate output on NVIDIA Jetson Orin Nano 8GB (Super)iPhone~38Simulate output on iPhone 1720 GB
fits · 17.0 GB free · up to 32K ctx
1 GPU
RX 7900 XT~378Simulate output on RX 7900 XT32–40 GB
fits · 29.0 GB free · up to 32K ctx
1 GPU · 3 Mac classes · 4 device families
RTX 5090~917Simulate output on RTX 5090M5 Max36 GB~254Simulate output on Mac Studio M5 Max (36 GB)M532 GB~85Simulate output on MacBook Air 13" M5 (32 GB)Orin~105Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~69Simulate 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 |
|---|---|---|---|---|
| LFM3 VL 2BLiquid AI | — | 2.3 GB | ~423 | same size |
| Qwen 1 8BAlibaba | — | 1.9 GB | ~542 | same size |
| Gemma 3n E2B IT Litert LmGoogle | — | 3.1 GB | ~496 | same size |
| Qwen3 1.7BAlibaba | — | 3.4 GB | ~374 | same size |
| MiniCPM 2B Sft BF16OpenBMB | — | 3 GB | ~291 | same size |
| Nemotron Labs Audex 2BNVIDIA | — | 6.1 GB | ~149 | same size |
Specifications
- Publisher
- vikhyatk
- Parameters
- 1.9B
- Architecture
- HfMoondream
- Release Date
- 2024-03-04
- License
- Apache 2.0
Frequently Asked Questions
How much VRAM does Moondream2 need?
Moondream2 requires 1.3 GB of VRAM at Q4_K_M, or 4.2 GB at BF16.
VRAM = Weights + KV Cache + Overhead
Weights = 1.9B × 4.8 bits ÷ 8 = 1.2 GB
KV Cache + Overhead ≈ 0.1 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.
VRAM usage by quantization
What's the best quantization for Moondream2?
For Moondream2, Q4_K_M (1.3 GB) offers the best balance of quality and VRAM usage. Q5_K_M (1.5 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 0.9 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run Moondream2 on a Mac?
Yes — MacBook Air 13" M3 (8 GB) and 145 other Macs can run Moondream2. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 1.3 GB of usable unified memory (RAM minus macOS overhead).
Can I run Moondream2 locally?
Yes — Moondream2 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Moondream2?
At Q4_K_M, Moondream2 can reach ~3780 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~516 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 ÷ 1.2700 × 0.65 = ~4094 tok/s
Estimated speed at Q4_K_M (1.3 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Moondream2?
At Q4_K_M, the download is about 1.16 GB. The full-precision BF16 version is 3.85 GB. The smallest option (Q2_K) is 0.82 GB.
Which GPUs can run Moondream2?
52 consumer GPUs can run Moondream2 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 Moondream2?
163 devices with unified memory can run Moondream2 at Q4_K_M (1.3 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.