h2oaiLlamaForCausalLM

H2o Danube3 500M Chat — Hardware Requirements & GPU Compatibility

Chat

H2o Danube3 500M Chat is a 514M-parameter open language model from h2oai. It supports a context window of up to 8,192 tokens. At Q4_K_M it needs about 0.71 GB of VRAM — see which GPUs and Macs can run it below.

514M params · dense8K contextApache 2.0Jul 2024146.3K downloads43 likes5.1K quant downloads

Your hardware

Verdicts below are for the reference RTX 4090. Set yours to personalise the page.

Detection reads your browser's GPU name with no data sent anywhere.

Runs great on a 24 GB GPU

Runs from 8 GB (Q4_K_M · 8K)Best on 24 GB (Q4_K_M · ~923 tok/s · 8K)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 H2o Danube3 500M Chat

One recommended way. Every other method, OS and quant is in the step-by-step guide.

Other ways

each opens step-by-step

Run in cloud

Fits on RTX 3060 12GB (11 GB headroom) · Q4_K_M

Generation speed: ~330 tok/s; generation speed
Cost per 1M output tokens: $0.070; per 1M output tokens

Deployon Vast.aiBest optionfrom $0.083/hr

Compare GPUs →

Run H2o Danube3 500M Chat

1.0 GB VRAM at 8K context · fits the RTX 4090 (23.0 GB free)

Works on macOS, Windows and Linux.

Install Ollama (Linux)
$curl -fsSL https://ollama.com/install.sh | sh
Install Ollama (macOS / Windows)
Download and run the app; it installs the ollama command.
Download Ollama ↗
Run Q4_K_M
$ollama run hf.co/bartowski/h2o-danube3-500m-chat-GGUF:Q4_K_M
Raise the context window
Type it inside the chat. For the API server instead, start it with OLLAMA_CONTEXT_LENGTH=8192 ollama serve.
$/set parameter num_ctx 8192
A larger context needs more memory than the weights alone.
Use the API
Ollama serves its API on http://localhost:11434.

Works on macOS, Windows and Linux.

Install LM Studio
Desktop app for macOS (Apple silicon), Windows and Linux.
lmstudio.ai ↗
Download the model
In the app press Ctrl/Cmd + Shift + M and paste https://huggingface.co/bartowski/h2o-danube3-500m-chat-GGUF, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get https://huggingface.co/bartowski/h2o-danube3-500m-chat-GGUF@Q4_K_M
Open on Hugging Face ↗
Load and chat
Open the model loader, select the model and set Context Length to 8192. Leave GPU offload at maximum if it fits.
Start the local server (optional)
$lms server start

Works on macOS, Windows and Linux.

Install llama.cpp (macOS)
MacPorts, Nix and conda-forge also work.
$brew install llama.cpp
Install options ↗
Install llama.cpp (Windows)
$winget install llama.cpp
Install llama.cpp (Linux)
Download the prebuilt llama-*-bin-ubuntu-* archive for your GPU backend (CPU, Vulkan) from the releases page, unpack it and run the binaries from that folder. Homebrew (brew install llama.cpp) and conda-forge also work on Linux.
llama.cpp releases ↗
Start the server
Downloads the model on first run. The chat UI and an OpenAI-compatible API are at http://localhost:8080.
$llama-server -hf bartowski/h2o-danube3-500m-chat-GGUF:Q4_K_M -c 8192 -ngl 99 --port 8080

Works on macOS, Windows and Linux.

Option A: Docker Model Runner
Docker Desktop: Settings > AI > Enable Docker Model Runner. Docker Engine on Linux: install the docker-model-plugin package.
$docker model run hf.co/bartowski/h2o-danube3-500m-chat-GGUF:Q4_K_M
Model Runner docs ↗
The quant tag after the repo is how Hugging Face cards write it; if your Docker version rejects it, drop the tag and pick the quant by repo.
Option B: llama.cpp server image (NVIDIA)
Needs the NVIDIA Container Toolkit. Models are cached in the llama-cache volume. Open http://localhost:8080.
$docker run --gpus all -p 8080:8080 -v llama-cache:/root/.cache ghcr.io/ggml-org/llama.cpp:server-cuda -hf bartowski/h2o-danube3-500m-chat-GGUF:Q4_K_M -c 8192 -ngl 99 --host 0.0.0.0 --port 8080
Image tags (cuda13, vulkan, rocm) ↗
Cache path inside the image is not in the docs; remove the -v flag if the volume complains.

Linux only.

Install vLLM
Linux only; on Windows run it inside WSL2. NVIDIA GPU shown; AMD ROCm, Intel and CPU builds have their own install pages.
$uv venv --python 3.12 --seed
$source .venv/bin/activate
$uv pip install vllm --torch-backend=auto
vLLM installation ↗
Serve the model
OpenAI-compatible API at http://localhost:8000/v1.
$vllm serve h2oai/h2o-danube3-500m-chat --max-model-len 8192
Runs the original weights, not the Q4_K_M GGUF; needs ~1 GB of GPU memory.

macOS only.

MLX supports only macOS, so follow these steps for that setup.

Install mlx-lm
$pip install mlx-lm
Chat in the terminal
$mlx_lm.chat --model mlx-community/h2o-danube3-500m-chat
Downloads the model on first run. The MLX repo has its own quantisation; it may not match the quant selected above.
Or start a local server
Listens on http://localhost:8080.
$mlx_lm.server --model mlx-community/h2o-danube3-500m-chat

Works on macOS, Windows and Linux.

Install Jan
jan.ai ↗
Find the model
Open the Hub tab, paste this model ID in Search, and choose Q4_K_M if offered.
$bartowski/h2o-danube3-500m-chat-GGUF
Use the local API (optional)
Settings > Local API Server > Start Server. It listens on http://127.0.0.1:1337.

Rent GPUs hourly to run Q4_K_M right there. Pricing comes from partner marketplaces and varies.

Available GPU rentals for Q4_K_M, prioritized by value
GPU	Speed	$/1M tokens	Rent
RTX 3060 12GBCheapestCheapest11 GB free	~330 tok/s	$0.070	Deploy on Vast.aifrom $0.083/hr (opens in new tab)	11 GB free · ~330 tok/s · $0.070 / 1M tokens
RTX 3070 (8 GB)7 GB free	~410 tok/s	$0.088	Deploy on RunPodfrom $0.13/hr (opens in new tab)	8 GB · 7 GB free · ~410 tok/s · $0.088 / 1M tokens
RTX 3070 (8 GB)7 GB free	~410 tok/s	$0.092	Deploy on Vast.aifrom $0.14/hr (opens in new tab)	8 GB · 7 GB free · ~410 tok/s · $0.092 / 1M tokens
RTX 3060 Ti (8 GB)7 GB free	~410 tok/s	$0.10	Deploy on Vast.aifrom $0.15/hr (opens in new tab)	8 GB · 7 GB free · ~410 tok/s · $0.10 / 1M tokens
RTX 5060 Ti 16GB15 GB free	~410 tok/s	$0.11	Deploy on Vast.aifrom $0.16/hr (opens in new tab)	15 GB free · ~410 tok/s · $0.11 / 1M tokens
RTX 3080 Ti (12 GB)11 GB free	~835 tok/s	$0.060	Deploy on RunPodfrom $0.18/hr (opens in new tab)	12 GB · 11 GB free · ~835 tok/s · $0.060 / 1M tokens
Tesla V100 PCIe 16GB15 GB free	~824 tok/s	$0.064	Deploy on RunPodfrom $0.19/hr (opens in new tab)	15 GB free · ~824 tok/s · $0.064 / 1M tokens
RTX 4090 (24 GB)23 GB free	~923 tok/s	$0.10	Deploy on RunPodfrom $0.34/hr (opens in new tab)	24 GB · 23 GB free · ~923 tok/s · $0.10 / 1M tokens

Compare GPUs →

How Much VRAM Does H2o Danube3 500M Chat Need?

Select a quantization level; the listed hardware adheres to it.

VRAM	+32K	+128K	File	Quality	On RTX 4090 (reference)
0.5 GB	—	—	~0.1 GB	2-bit · i-quant extreme compression — Importance-weighted 2-bit, extreme compression — significant quality loss	Fits: 8K context · ~1213 tok/s
0.6 GB	—	—	0.2 GB	2-bit · i-quant extra small — Importance-weighted 2-bit, extra small	Fits: 8K context · ~1191 tok/s
0.6 GB	—	—	0.2 GB	2-bit · i-quant small — Importance-weighted 2-bit, small	Fits: 8K context · ~1170 tok/s
0.6 GB	—	—	0.2 GB	2-bit · i-quant medium — Importance-weighted 2-bit, medium	Fits: 8K context · ~1150 tok/s
0.6 GB	—	—	0.2 GB	3-bit · i-quant — Importance-weighted 3-bit	Fits: 8K context · ~1092 tok/s
0.6 GB	—	—	0.2 GB	3-bit · i-quant extra small — Importance-weighted 3-bit, extra small	Fits: 8K context · ~1074 tok/s
0.6 GB	—	—	0.2 GB	3-bit · i-quant small — Importance-weighted 3-bit, small	Fits: 8K context · ~1057 tok/s
0.6 GB	—	—	~0.2 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Fits: 8K context · ~1057 tok/s
0.6 GB	—	—	0.2 GB	3-bit small — 3-bit small quantization	Fits: 8K context · ~1040 tok/s
0.6 GB	—	—	0.2 GB	3-bit · i-quant medium — Importance-weighted 3-bit, medium	Fits: 8K context · ~1040 tok/s
0.7 GB	—	—	0.3 GB	3-bit medium — 3-bit medium quantization	Fits: 8K context · ~1008 tok/s
0.7 GB	—	—	0.3 GB	3-bit large — 3-bit large quantization	Fits: 8K context · ~993 tok/s
0.7 GB	—	—	0.3 GB	4-bit · legacy — 4-bit legacy quantization	Fits: 8K context · ~993 tok/s
0.7 GB	—	—	0.3 GB	4-bit · i-quant compact — Importance-weighted 4-bit, compact	Fits: 8K context · ~964 tok/s
0.7 GB	—	—	0.3 GB	4-bit small — 4-bit small quantization	Fits: 8K context · ~950 tok/s
0.7 GB	—	—	0.3 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Fits: 8K context · ~923 tok/s
0.7 GB	—	—	0.4 GB	4-bit large — 4-bit large quantization	Fits: 8K context · ~910 tok/s
0.8 GB	—	—	0.4 GB	5-bit small — 5-bit small quantization	Fits: 8K context · ~874 tok/s
0.8 GB	—	—	0.4 GB	5-bit large — 5-bit large quantization	Fits: 8K context · ~851 tok/s
0.8 GB	—	—	0.4 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Fits: 8K context · ~851 tok/s
0.8 GB	—	—	0.4 GB	6-bit, very good — 6-bit quantization, very good quality	Fits: 8K context · ~799 tok/s
0.9 GB	—	—	0.5 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Fits: 8K context · ~720 tok/s
1.4 GB	—	—	~1 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Fits: 8K context · ~458 tok/s

est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

1.0GBat 8K context

Weights: 0.3 GB
KV cache: 0.4 GB
Overhead: 0.3 GB

Fits RTX 4090 (reference) · 23.0 GB free · ~923 tok/s

≥50 tok/s needs ≥55 GB/s · RTX 5090 ~1641, RTX 3090 Ti ~923

Which Hardware Can Run H2o Danube3 500M Chat?

Q4_K_M · 0.7 GB

H2o Danube3 500M Chat (Q4_K_M) requires 0.7 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 1+ GB is recommended. Using the full 8K context window can add up to 0.3 GB, bringing total usage to 1.0 GB. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 163 devices with unified memory can run H2o Danube3 500M Chat, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M

Typical hardwareQuantVRAMtok/sContext that fits

8 GB · RTX 4060Q4_K_M · 1 GB · ~249 tok/s · 8K1~2498K8 GB tier →

12 GB · RTX 3060 12GBQ4_K_M · 1 GB · ~330 tok/s · 8K1~3308K12 GB tier →

16 GB · RTX 5060 Ti 16GBQ4_K_M · 1 GB · ~410 tok/s · 8K1~4108K16 GB tier →

24 GB · RTX 4090referenceQ4_K_M · 1 GB · ~923 tok/s · 8K1~9238K24 GB tier →

Mac 32 GB · Mac Mini M4Q4_K_M · 1 GB · ~118 tok/s · 8K1~1188KMacs →

Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 1 GB · ~538 tok/s · 8K1~5388KMacs →

Alternatives to H2o Danube3 500M Chat

Similar models

4tok/s on RTX 4090

Model	Score	VRAM	tok/s	Why it is similar
Falcon H1 0.5B BaseTII UAE	—	1.2 GB	~950	same size
Qwen2.5 0.5B InstructAlibaba	—	0.8 GB	~1057	same size
Qwen2 0.5B InstructAlibaba	—	0.8 GB	~1057	same size
Qwen3 4B DFlash B16z-lab	—	0.8 GB	~1008	same size

Specifications

Publisher: h2oai
Parameters: 514M
Architecture: LlamaForCausalLM
Context Length: 8,192 tokens
Vocabulary Size: 32,000
Release Date: 2024-07-04
License: Apache 2.0

Resources

Hugging Face model card (opens in a new tab)

Frequently Asked Questions

How much VRAM does H2o Danube3 500M Chat need?

H2o Danube3 500M Chat requires 0.7 GB of VRAM at Q4_K_M, or 1.4 GB at BF16.

VRAM = Weights + KV Cache + Overhead

Weights = 514M × 4.8 bits ÷ 8 = 0.3 GB

KV Cache + Overhead ≈ 0.4 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 ≈ 0.7 GB (at full 8K context)

VRAM usage by quantization

Q4_K_M

0.7 GB

Q4_K_M + full context

1.0 GB

Learn more about VRAM estimation →

What's the best quantization for H2o Danube3 500M Chat?

For H2o Danube3 500M Chat, Q4_K_M (0.7 GB) offers the best balance of quality and VRAM usage. Q4_K_L (0.7 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 0.5 GB.

VRAM requirement by quantization

IQ2_XXS

0.5 GB~53%

IQ3_S

0.6 GB~75%

Q3_K_L

0.7 GB~86%

Q4_K_M ★

0.7 GB~89%

Q5_K_S

0.8 GB~92%

BF16

1.4 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run H2o Danube3 500M Chat on a Mac?

Yes — MacBook Air 13" M3 (8 GB) and 145 other Macs can run H2o Danube3 500M Chat. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 0.7 GB of usable unified memory (RAM minus macOS overhead).

Can I run H2o Danube3 500M Chat locally?

Yes — H2o Danube3 500M Chat can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is H2o Danube3 500M Chat?

At Q4_K_M, H2o Danube3 500M Chat can reach ~6761 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~923 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 ÷ 0.71 × 0.65 = ~7324 tok/s

Estimated speed at Q4_K_M (0.7 GB)

NVIDIA B200

~7324 tok/s

NVIDIA GeForce RTX 4090

~923 tok/s

NVIDIA B300

~7324 tok/s

AMD Instinct MI350X

~6761 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of H2o Danube3 500M Chat?

At Q4_K_M, the download is about 0.31 GB. The full-precision BF16 version is 1.03 GB. The smallest option (IQ2_XXS) is 0.14 GB.

Which GPUs can run H2o Danube3 500M Chat?

52 consumer GPUs can run H2o Danube3 500M Chat 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 H2o Danube3 500M Chat?

163 devices with unified memory can run H2o Danube3 500M Chat at Q4_K_M (0.7 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.