Skip to content
localmodel.run

Text model · Sarvam

S Sarvam-30B: RAM and VRAM requirements

Sarvam-30B needs about 21.7 GB to run at Q4_K_M (the Q4_K_M GGUF file is ~19.6 GB to download; KV cache and overhead add the rest), or about 34 GB at Q8_0. The lightest hardware that runs it is Nvidia GeForce RTX 4090 (24GB).

Sarvam family · 30B params (Mixture-of-Experts: activates only 2.4B of 30B params per token, so generation is faster than the total size suggests) · released Mar 2026.

Shopping for hardware? See what runs Sarvam-30B →

License Apache-2.0 · Commercial OK ↓ 35.2K/mo ♥ 218 on HuggingFace
Q4_K_M GGUF
19.6 GB
Q8_0 GGUF
-
Memory @ Q4 (4k)
~21.7 GB
Context
64 k

Q4_K_M and Q8_0 are GGUF download (file) sizes. Memory @ Q4 includes the KV cache and overhead, so it runs larger than the file on disk.

Will it run on your device?

Sarvam-30B runs on 12 of 40 tracked devices at Q4_K_M.

9 run well 3 tight fit 28 too small
Fit check Q4_K_M
No, not enough memory
needs 21.7 GB usable 10.5 GB
See the full breakdown

Memory breakdown

Weights (Q4_K_M)19.6 GB
+
KV cache (4k)1.3 GB
+
Overhead0.8 GB
=
Total21.7 GB

How context length changes it

4k context ~21.7 GB 32k context ~30.9 GB 128k context ~62.5 GB

Longer context grows the KV cache, which for Sarvam-30B is sized by its 2.4B active params, not the full 30B. It needs ~21.7 GB at 4k and ~62.5 GB at 128k.

Quantization sizes

GGUF quantson disk
Quantization Size on disk
Q2_K 12.6 GB est
Q3_K_M 14.7 GB est
Q4_K_M (default) 19.6 GB
Q5_K_M 21.4 GB est
Q6_K 24.6 GB est
Q8_0 31.9 GB est
FP16 60 GB est

Lower quant = smaller and faster, slightly lower quality. Q4_K_M is the common default.

Run it

llama.cpp
$ llama-cli -hf sarvamai/sarvam-30b-gguf:Q4_K_M
LM Studio
$ lms get sarvamai/sarvam-30b-gguf

Which devices can run Sarvam-30B?

Apple Silicon Macs

RAM-only laptops

iPhone & iPad

Android

NVIDIA GPUs

AMD GPUs

Same job, different size

Pick by what fits your memory: step down to free up VRAM, or step up if you have headroom.

Next size down Granite 4.1 30B 28.9B Next size up Qwen3 30B-A3B 30.5B

Similar models

Head-to-head

FAQ

How much VRAM or RAM does Sarvam-30B need?

At Q4_K_M, Sarvam-30B needs about 21.7 GB (weights ~19.6 GB + KV cache + overhead) at a 4k context. At Q8_0 budget ~34 GB.

What is the Q4_K_M GGUF file size of Sarvam-30B?

The Q4_K_M GGUF file is about 19.6 GB to download, and the Q8_0 GGUF is about 31.9 GB. That is the weights file on disk; to run it you also need room for the KV cache and overhead, so budget ~21.7 GB of memory at Q4_K_M.

Can Sarvam-30B run on a laptop?

Sarvam-30B is large; you need a 24 GB+ GPU or a 32-48 GB Mac at Q4_K_M.

Is Sarvam-30B cheaper to run because it is a MoE model?

It is faster, not lighter. Sarvam-30B activates only 2.4B of 30B params per token (so it runs quickly), but all experts must stay in memory, so it still needs memory for the full 30B.

Can I use Sarvam-30B commercially?

Yes. Sarvam-30B is licensed Apache-2.0, which permits commercial use.

Understand the numbers

Short guides to the ideas behind Sarvam-30B's memory and quant figures.

MoE with 128 sparse experts, top-6 routing, 2.4B active params. Released 2026-03 under Apache 2.0. Q4_K_M 19.6GB confirmed by summing the 6 shards in the official GGUF repo. No official Q8_0 GGUF released.

Sources

Last validated 2026-08-03. Memory figures are estimates. See methodology.