Skip to content
localmodel.run

Text model · Sarvam

S Sarvam-1 2B: RAM and VRAM requirements

Sarvam-1 2B needs about 2.7 GB to run at Q4_K_M (the Q4_K_M GGUF file is ~1.55 GB to download; KV cache and overhead add the rest), or about 3.8 GB at Q8_0. The lightest hardware that runs it is Apple M1 (8GB).

Sarvam family · 2B params · released Oct 2024.

Shopping for hardware? See what runs Sarvam-1 2B →

License Sarvam non-commercial · Non-commercial ↓ 8.1K/mo ♥ 142 on HuggingFace
Q4_K_M GGUF
1.55 GB
Q8_0 GGUF
2.69 GB
Memory @ Q4 (4k)
~2.7 GB
Context
8 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-1 2B runs on 40 of 40 tracked devices at Q4_K_M.

40 run well 0 tight fit 0 too small
Fit check Q4_K_M
Yes, it runs fast
needs 2.7 GB usable 10.5 GB
See the full breakdown

Memory breakdown

Weights (Q4_K_M)1.55 GB
+
KV cache (4k)0.3 GB
+
Overhead0.8 GB
=
Total2.7 GB

How context length changes it

4k context ~2.7 GB 32k context ~5.1 GB 128k context ~13.3 GB

Sarvam-1 2B is small enough that the context window is the thing to size for: 128k needs ~13.3 GB versus ~2.7 GB at 4k, so pick the window you actually use.

Speed drops too: every token re-reads the KV cache, so at 128k context Sarvam-1 2B generates at roughly ~12% of its short-context speed. How this is estimated.

Quantization sizes

GGUF quantson disk
Quantization Size on disk
Q2_K 0.8 GB est
Q3_K_M 1 GB est
Q4_K_M (default) 1.55 GB
Q5_K_M 1.4 GB est
Q6_K 1.6 GB est
Q8_0 2.69 GB
FP16 5.05 GB

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

Run it

llama.cpp
$ llama-cli -hf bartowski/sarvam-1-GGUF:Q4_K_M
LM Studio
$ lms get bartowski/sarvam-1-GGUF

Which devices can run Sarvam-1 2B?

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 Qwen3 1.7B 1.7B Next size up Granite 3.1 2B 2.53B

Similar models

Head-to-head

FAQ

How much VRAM or RAM does Sarvam-1 2B need?

At Q4_K_M, Sarvam-1 2B needs about 2.7 GB (weights ~1.55 GB + KV cache + overhead) at a 4k context. At Q8_0 budget ~3.8 GB.

What is the Q4_K_M GGUF file size of Sarvam-1 2B?

The Q4_K_M GGUF file is about 1.55 GB to download, and the Q8_0 GGUF is about 2.69 GB. That is the weights file on disk; to run it you also need room for the KV cache and overhead, so budget ~2.7 GB of memory at Q4_K_M.

Can Sarvam-1 2B run on a laptop?

Yes, Sarvam-1 2B fits on a 16 GB machine at Q4_K_M and runs on Apple Silicon or a 12 GB+ GPU comfortably.

Can I use Sarvam-1 2B commercially?

No. Non-commercial use only per the Sarvam license. Check the current HuggingFace model card for updates.

Understand the numbers

Short guides to the ideas behind Sarvam-1 2B's memory and quant figures.

2B dense model trained from scratch, optimized for 10 Indic languages plus English. Released 2024-10-24 under the Sarvam non-commercial license (not Apache). GGUF sizes from bartowski: Q4_K_M 1.55GB, Q8_0 2.69GB. Context 8K from the model card.

Sources

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