ricursiveAIRicursiveForCausalLM

Ripple 65B Preview — Hardware Requirements & GPU Compatibility

ChatFunctions

Ripple 65B Preview is a 64.8B-parameter open language model from ricursiveAI. It supports a context window of up to 2,097,152 tokens. At Q4_K_M it needs about 39.51 GB of VRAM — see which GPUs and Macs can run it below.

64.8B params · dense2097K contextApache 2.0Nov 20268.5K downloads575 likes103.8K 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.

B

Decent on a 24 GB GPU

Runs on 24 GB (IQ2_S · ~31 tok/s · 16K)Macs from 32 GB (IQ3_XXS · ~11 tok/s)

Apache 2.0

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Ripple 65B Preview

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

one command · needs Ollama installed

$ollama run hf.co/AtomicChat/Ripple-65B-Preview-GGUF:Q4_K_M

Q4_K_M · needs ~41.9 GB · 16K context

Download: Ripple-65B-Preview-Q4_K_M.gguf38.80 GBAtomicChat

install, OS, other quants

Ways to runOther ways

each opens step-by-step

Run in cloud

Runs on A100 80GB (38 GB headroom) · Q4_K_M

Generation speed: ~34 tok/s; generation speed
Cost per 1M output tokens: $9.38; per 1M output tokens

Deployon Vast.aiBest valuefrom $1.13/hr

Compare GPUs →

or

How Much VRAM Does Ripple 65B Preview Need?

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

Quantization	VRAM	+32K	+128K	File	Quality	On RTX 4090 (reference)
	18.4 GB	23.5 GB	39.6 GB	20 GB	2-bit · i-quant extreme compression — Importance-weighted 2-bit, extreme compression — significant quality loss	Fits: 32K context · ~36 tok/s
	20.1 GB	25.1 GB	41.2 GB	22.1 GB	2-bit · i-quant extra small — Importance-weighted 2-bit, extra small	Fits: 16K context · ~33 tok/s
	20.9 GB	25.9 GB	42 GB	25.3 GB	2-bit · i-quant small — Importance-weighted 2-bit, small	Fits: 16K context · ~31 tok/s
	25.7 GB	30.8 GB	46.9 GB	28 GB	3-bit · i-quant — Importance-weighted 3-bit	Does not fit: needs 28.1 GB
	28.2 GB	33.2 GB	49.3 GB	32.4 GB	3-bit · i-quant small — Importance-weighted 3-bit, small	Does not fit: needs 30.5 GB
	28.2 GB	33.2 GB	49.3 GB	~27.5 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Does not fit: needs 30.5 GB
	32.2 GB	37.3 GB	53.4 GB	~31.6 GB	3-bit medium — 3-bit medium quantization	Does not fit: needs 34.6 GB
	35.5 GB	40.5 GB	56.6 GB	38.7 GB	4-bit · i-quant compact — Importance-weighted 4-bit, compact	Does not fit: needs 37.8 GB
	39.5 GB	44.5 GB	60.7 GB	38.8 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Does not fit: needs 41.9 GB
	46.8 GB	51.8 GB	67.9 GB	47.9 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Does not fit: needs 49.1 GB
	54.1 GB	59.1 GB	75.2 GB	57.7 GB	6-bit, very good — 6-bit quantization, very good quality	Does not fit: needs 56.4 GB
	65.4 GB	70.5 GB	86.6 GB	68.8 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Does not fit: needs 67.8 GB
	130.2 GB	135.2 GB	151.3 GB	129.6 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Does not fit: needs 132.6 GB

est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

41.9GBat 16K context

Weights: 38.9 GB
KV cache: 2.7 GB
Overhead: 0.3 GB

Doesn't fit RTX 4090 (reference) · needs 17.9 GB more, IQ2_S fits

≥10 tok/s needs ≥605 GB/s · M3 Ultra ~15, M2 Ultra ~14

Or .

Which Hardware Can Run Ripple 65B Preview?

Q4_K_M · 39.5 GB

Ripple 65B Preview (Q4_K_M) requires 39.5 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 52+ GB is recommended. Using the full 2097K context window can add up to 343.3 GB, bringing total usage to 382.8 GB. No consumer GPU has enough memory. 67 devices with unified memory can run Ripple 65B Preview, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M5 Max (48 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M

Typical hardwareQuantVRAMtok/sContext that fits

8–16 GB · RTX 4060, RTX 3060 12GB, RTX 5060 Ti 16GBDoesn't fit — smallest quant needs 20.8 GB8 GB tier →12 GB tier →16 GB tier →

24 GB · RTX 4090referenceIQ2_S · 23.2 GB · ~31 tok/s · 16K23.2~3116K24 GB tier →

Mac 32 GB · Mac Mini M4IQ3_XXS · 28.1 GB · ~3 tok/s · 16K28.1~316KMacs →

Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 41.9 GB · ~10 tok/s · 64K41.9~1064KMacs →

Alternatives to Ripple 65B Preview

Similar models

3tok/s on RTX 4090

Model	Score	VRAM	tok/s	Why it is similar
CodeLlama 70B Instruct HFMeta	—	23.2 GB	~29	same size
Mixtral 34Bx2 MoE 60Bcloudyu	—	40.8 GB	—needs > 24 GB	same size
H2ogpt 4096 Llama2 70B Chath2oai	—	43 GB	—needs > 24 GB	same size

Specifications

Publisher: ricursiveAI
Parameters: 64.8B
Architecture: RicursiveForCausalLM
Context Length: 2,097,152 tokens
Vocabulary Size: 100,352
Release Date: 2026-11-04
License: Apache 2.0

Resources

Hugging Face model card (opens in a new tab)

Frequently Asked Questions

How much VRAM does Ripple 65B Preview need?

Ripple 65B Preview requires 39.5 GB of VRAM at Q4_K_M, or 130.2 GB at BF16. Full 2097K context adds up to 343.3 GB (382.8 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 64.8B × 4.8 bits ÷ 8 = 38.9 GB

KV Cache + Overhead ≈ 0.6 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 ≈ 343.9 GB (at full 2097K context)

VRAM usage by quantization

Q4_K_M

39.5 GB

Q4_K_M + full context

382.8 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 4090 run Ripple 65B Preview?

Yes, at IQ2_S (20.9 GB) or lower. Higher quantizations like IQ3_XXS (25.7 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.

What's the best quantization for Ripple 65B Preview?

For Ripple 65B Preview, Q4_K_M (39.5 GB) offers the best balance of quality and VRAM usage. Q5_K_M (46.8 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 18.4 GB.

VRAM requirement by quantization

IQ2_XXS

18.4 GB~53%

IQ3_XXS

25.7 GB~70%

Q3_K_M

32.2 GB~83%

Q4_K_M ★

39.5 GB~89%

Q5_K_M

46.8 GB~92%

BF16

130.2 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Ripple 65B Preview on a Mac?

Yes — Mac Studio M1 Max (32 GB) and 84 other Macs can run Ripple 65B Preview, the smallest at IQ3_XXS. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 39.5 GB of usable unified memory (RAM minus macOS overhead), from Mac Studio M5 Max (48 GB).

Can I run Ripple 65B Preview locally?

Yes — Ripple 65B Preview can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Ripple 65B Preview?

At Q4_K_M, Ripple 65B Preview can reach ~122 tok/s on AMD Instinct MI350X. 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 ÷ 39.5 × 0.65 = ~132 tok/s

Estimated speed at Q4_K_M (39.5 GB)

NVIDIA B200

~132 tok/s

NVIDIA B300

~132 tok/s

AMD Instinct MI350X

~122 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 Ripple 65B Preview?

At Q4_K_M, the download is about 38.87 GB. The full-precision BF16 version is 129.57 GB. The smallest option (IQ2_XXS) is 17.82 GB.

Which GPUs can run Ripple 65B Preview?

No single consumer GPU has enough VRAM to run Ripple 65B Preview at Q4_K_M (39.5 GB). Multi-GPU or professional hardware is required.

Which devices can run Ripple 65B Preview?

67 devices with unified memory can run Ripple 65B Preview at Q4_K_M (39.5 GB), including ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB), GMKtec EVO-X2 (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.

Run Ripple 65B Preview

41.9 GB VRAM at 16K context

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/AtomicChat/Ripple-65B-Preview-GGUF:Q4_K_M
Raise the context window
Type it inside the chat. For the API server instead, start it with OLLAMA_CONTEXT_LENGTH=16384 ollama serve.
$/set parameter num_ctx 16384
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/AtomicChat/Ripple-65B-Preview-GGUF, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get https://huggingface.co/AtomicChat/Ripple-65B-Preview-GGUF@Q4_K_M
Open on Hugging Face ↗
Load and chat
Open the model loader, select the model and set Context Length to 16384. 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 AtomicChat/Ripple-65B-Preview-GGUF:Q4_K_M -c 16384 -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/AtomicChat/Ripple-65B-Preview-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 AtomicChat/Ripple-65B-Preview-GGUF:Q4_K_M -c 16384 -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 ricursiveAI/Ripple-65B-Preview --max-model-len 16384
Runs the original weights, not the Q4_K_M GGUF; needs ~143 GB of GPU memory.

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.
$AtomicChat/Ripple-65B-Preview-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
A100 80GBCheapestCheapest38 GB free	~34 tok/s	$9.38	Deploy on Vast.aifrom $1.13/hr (opens in new tab)	38 GB free · ~34 tok/s · $9.38 / 1M tokens
RTX PRO 6000 Max-Q (96 GB)54 GB free	~30 tok/s	$10.71	Deploy on Vast.aifrom $1.14/hr (opens in new tab)	96 GB · 54 GB free · ~30 tok/s · $10.71 / 1M tokens
A100 80GB38 GB free	~34 tok/s	$11.53	Deploy on RunPodfrom $1.39/hr (opens in new tab)	38 GB free · ~34 tok/s · $11.53 / 1M tokens
RTX PRO 6000 (96 GB)54 GB free	~30 tok/s	$13.62	Deploy on Vast.aifrom $1.45/hr (opens in new tab)	96 GB · 54 GB free · ~30 tok/s · $13.62 / 1M tokens
RTX PRO 6000 Server (96 GB)54 GB free	~26 tok/s	$15.75	Deploy on Vast.aifrom $1.49/hr (opens in new tab)	96 GB · 54 GB free · ~26 tok/s · $15.75 / 1M tokens
RTX PRO 6000 (96 GB)54 GB free	~30 tok/s	$15.91	Deploy on RunPodfrom $1.69/hr (opens in new tab)	96 GB · 54 GB free · ~30 tok/s · $15.91 / 1M tokens
RTX PRO 6000 Server (96 GB)54 GB free	~26 tok/s	$17.85	Deploy on RunPodfrom $1.69/hr (opens in new tab)	96 GB · 54 GB free · ~26 tok/s · $17.85 / 1M tokens
H100 (80 GB)38 GB free	~34 tok/s	$16.50	Deploy on RunPodfrom $1.99/hr (opens in new tab)	80 GB · 38 GB free · ~34 tok/s · $16.50 / 1M tokens

Compare GPUs →