aoxoSarvamMoEForCausalLM

Sarvam 30B Uncensored — Hardware Requirements & GPU Compatibility

Chat

Sarvam 30B Uncensored is a 32.2B-parameter open language model from aoxo. It supports a context window of up to 131,072 tokens. At Q4_K_M it needs about 19.63 GB of VRAM — see which GPUs and Macs can run it below.

32.2B params · MoE4.5B active131K contextApache 2.0Mar 2026413 downloads6 likes152 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.

S

Runs great on a 24 GB GPU

Runs from 16 GB (Q2_K · 64K)Best on 24 GB (Q4_K_M · ~190 tok/s · 128K)Macs from 18 GB (Q2_K · ~44 tok/s)

Apache 2.0

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Sarvam 30B Uncensored

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

Recommended

Ollama

one command · needs Ollama installed

$ollama run hf.co/ThatCultivator/sarvam-30b-Uncensored-gguf:Q4_K_M

Q4_K_M fits fully on the RTX 4090 · 4.1 GB free · 16K context

Download: sarvam-30b-uncensored.Q4_K_M.gguf19.58 GBThatCultivator

install, OS, other quants

Other ways

each opens step-by-step

Run in cloud

Fits on RTX 3090 (24 GB) (4 GB headroom) · Q4_K_M

Generation speed: ~179 tok/s; generation speed
Cost per 1M output tokens: $0.33; per 1M output tokens

Deployon Vast.aiBest optionfrom $0.22/hr

Compare GPUs →

or

How Much VRAM Does Sarvam 30B Uncensored Need?

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

Quantization	VRAM	+32K	+128K	File	Quality	On RTX 4090 (reference)
	14 GB	14.6 GB	16.5 GB	~13.7 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Fits: 128K context · ~247 tok/s
	16.0 GB	16.6 GB	18.5 GB	15.7 GB	3-bit medium — 3-bit medium quantization	Fits: 128K context · ~223 tok/s
	19.6 GB	20.2 GB	22.1 GB	19.6 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Fits: 128K context · ~190 tok/s
	23.3 GB	23.9 GB	25.8 GB	~22.9 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Fits: 32K context · ~166 tok/s
	26.9 GB	27.5 GB	29.4 GB	26.4 GB	6-bit, very good — 6-bit quantization, very good quality	Does not fit: needs 27.2 GB
	32.5 GB	33.1 GB	35 GB	~32.2 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Does not fit: needs 32.8 GB
	64.7 GB	65.3 GB	67.2 GB	~64.3 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Does not fit: needs 64.9 GB

est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

19.9GBat 16K context

Weights: 19.3 GB
KV cache: 0.3 GB
Overhead: 0.3 GB

Fits RTX 4090 (reference) · 4.1 GB free · ~190 tok/s

≥50 tok/s needs ≥224 GB/s · RTX 5090 ~293, RTX 3090 Ti ~190

Which Hardware Can Run Sarvam 30B Uncensored?

Q4_K_M · 19.6 GB

Sarvam 30B Uncensored (Q4_K_M) requires 19.6 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 26+ GB is recommended. Using the full 131K context window can add up to 2.5 GB, bringing total usage to 22.1 GB. 8 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 122 devices with unified memory can run Sarvam 30B Uncensored, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M1 Max (32 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M

Typical hardwareQuantVRAMtok/sContext that fits

8–12 GB · RTX 4060, RTX 3060 12GBDoesn't fit — smallest quant needs 14.3 GB8 GB tier →12 GB tier →

16 GB · RTX 5060 Ti 16GBQ2_K · 14.3 GB · ~128 tok/s · 64K14.3~12864K16 GB tier →

24 GB · RTX 4090referenceQ4_K_M · 19.9 GB · ~190 tok/s · 128K19.9~190128K24 GB tier →

Mac 32 GB · Mac Mini M4Q4_K_M · 19.9 GB · ~27 tok/s · 128K19.9~27128KMacs →

Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 19.9 GB · ~91 tok/s · 128K19.9~91128KMacs →

Alternatives to Sarvam 30B Uncensored

Similar models

6tok/s on RTX 4090

Model	Score	VRAM	tok/s	Why it is similar
Hy Memory 30B A3BTencent	—	19.1 GB	~184	same size
Muse Glimmer 30BMeta	—	19.6 GB	~36	same size
NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA	—	19.3 GB	~197	same size
Hy MT2 30B A3BTencent	—	19.1 GB	~170	same size
EXAONE 4.5 33BLGAI-EXAONE	—	23.9 GB	~33	same size
Gemma 4 31BGoogle	—	22.6 GB	~31	same size

Specifications

Publisher: aoxo
Parameters: 32.2B
Active parameters: 4.5B per token
Architecture: SarvamMoEForCausalLM
Context Length: 131,072 tokens
Vocabulary Size: 262,144
Release Date: 2026-03-07
License: Apache 2.0

Resources

Hugging Face model card (opens in a new tab)

Frequently Asked Questions

How much VRAM does Sarvam 30B Uncensored need?

Sarvam 30B Uncensored requires 19.6 GB of VRAM at Q4_K_M, or 64.7 GB at BF16. Full 131K context adds up to 2.5 GB (22.1 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 32.2B × 4.8 bits ÷ 8 = 19.3 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 ≈ 2.8 GB (at full 131K context)

VRAM usage by quantization

Q4_K_M

19.6 GB

Q4_K_M + full context

22.1 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 4090 run Sarvam 30B Uncensored?

Yes, at Q5_K_M (23.3 GB) or lower. Higher quantizations like Q6_K (26.9 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.

What's the best quantization for Sarvam 30B Uncensored?

For Sarvam 30B Uncensored, Q4_K_M (19.6 GB) offers the best balance of quality and VRAM usage. Q5_K_M (23.3 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 14 GB.

VRAM requirement by quantization

Q2_K

14.0 GB~75%

Q4_K_M ★

19.6 GB~89%

Q5_K_M

23.3 GB~92%

Q6_K

26.9 GB~95%

Q8_0

32.5 GB~99%

BF16

64.7 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Sarvam 30B Uncensored on a Mac?

Yes — MacBook Pro 14" M3 Pro (18 GB) and 108 other Macs can run Sarvam 30B Uncensored, the smallest at Q2_K. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 19.6 GB of usable unified memory (RAM minus macOS overhead), from Mac Mini M5 Pro (24 GB).

Can I run Sarvam 30B Uncensored locally?

Yes — Sarvam 30B Uncensored can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Sarvam 30B Uncensored?

At Q4_K_M, Sarvam 30B Uncensored can reach ~229 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~190 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 → 2.8 GB active ÷ (8000 × 0.65) = 0.531 ms, plus 19 layers × 0.055 ms = 1.045 ms, so 1000 ÷ 1.576 ms = ~635 tok/s

Estimated speed at Q4_K_M (19.6 GB)

NVIDIA B200

~635 tok/s

NVIDIA GeForce RTX 4090

~190 tok/s

NVIDIA B300

~635 tok/s

NVIDIA GH200 Grace Hopper Superchip

~523 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 Sarvam 30B Uncensored?

At Q4_K_M, the download is about 19.29 GB. The full-precision BF16 version is 64.31 GB. The smallest option (Q2_K) is 13.66 GB.

Which GPUs can run Sarvam 30B Uncensored?

8 consumer GPUs can run Sarvam 30B Uncensored at Q4_K_M. Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XT, AMD Radeon RX 7900 XTX. 1 GPU has plenty of headroom for comfortable inference.

Which devices can run Sarvam 30B Uncensored?

122 devices with unified memory can run Sarvam 30B Uncensored at Q4_K_M (19.6 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.

Run Sarvam 30B Uncensored

19.9 GB VRAM at 16K context · fits the RTX 4090 (4.1 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/ThatCultivator/sarvam-30b-Uncensored-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/ThatCultivator/sarvam-30b-Uncensored-gguf, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get https://huggingface.co/ThatCultivator/sarvam-30b-Uncensored-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 ThatCultivator/sarvam-30b-Uncensored-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/ThatCultivator/sarvam-30b-Uncensored-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 ThatCultivator/sarvam-30b-Uncensored-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 aoxo/sarvam-30b-uncensored --max-model-len 16384
Runs the original weights, not the Q4_K_M GGUF; needs ~71 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.
$ThatCultivator/sarvam-30b-Uncensored-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 3090 (24 GB)CheapestCheapest4 GB free	~179 tok/s	$0.33	Deploy on Vast.aifrom $0.22/hr (opens in new tab)	24 GB · 4 GB free · ~179 tok/s · $0.33 / 1M tokens
RTX A5000 (24 GB)4 GB free	~152 tok/s	$0.42	Deploy on Vast.aifrom $0.23/hr (opens in new tab)	24 GB · 4 GB free · ~152 tok/s · $0.42 / 1M tokens
TITAN RTX (24 GB)4 GB free	~136 tok/s	$0.50	Deploy on Vast.aifrom $0.24/hr (opens in new tab)	24 GB · 4 GB free · ~136 tok/s · $0.50 / 1M tokens
RTX PRO 4000 (24 GB)4 GB free	~136 tok/s	$0.68	Deploy on Vast.aifrom $0.33/hr (opens in new tab)	24 GB · 4 GB free · ~136 tok/s · $0.68 / 1M tokens
RTX PRO 4500 (32 GB)12 GB free	~173 tok/s	$0.55	Deploy on RunPodfrom $0.34/hr (opens in new tab)	32 GB · 12 GB free · ~173 tok/s · $0.55 / 1M tokens
RTX 4090 (24 GB)4 GB free	~190 tok/s	$0.50	Deploy on RunPodfrom $0.34/hr (opens in new tab)	24 GB · 4 GB free · ~190 tok/s · $0.50 / 1M tokens
A40 (48 GB)28 GB free	~140 tok/s	$0.69	Deploy on RunPodfrom $0.35/hr (opens in new tab)	48 GB · 28 GB free · ~140 tok/s · $0.69 / 1M tokens
L4 (24 GB)4 GB free	~66 tok/s	$1.86	Deploy on RunPodfrom $0.44/hr (opens in new tab)	24 GB · 4 GB free · ~66 tok/s · $1.86 / 1M tokens

Compare GPUs →