text model · Sarvam · Windows
Can I run Sarvam-M 24B on Nvidia GeForce RTX 4090 (24GB)?
Yes. Sarvam-M 24B runs on Nvidia GeForce RTX 4090 (24GB) at Q4_K_M (~16.3 GB of ~23 GB usable).
Runs at Q4_K_M using ~16.3 GB of ~23 GB usable.
That figure is at a 4k context and moves about ±15% as context length changes. Nvidia GeForce RTX 4090 (24GB) leaves ~6.7 GB of headroom.
- Q4_K_M needed
- ~16.3 GB
- Usable on device
- ~23 GB
- Device memory
- 24 GB
- Best quant
- Q4_K_M
Which quant fits
Running cost · estimate
- Power draw
- ~450 W
- Electricity / 1M tokens
- ~$0.41
- Pays for itself after
- ~17,767M tok
At ~$0.15/kWh and the estimated ~46 tok/s, a million generated tokens costs about $0.41 in electricity. Versus a hosted API at ~$0.5 per million tokens, the ~$1,599 Nvidia GeForce RTX 4090 (24GB) pays for itself after roughly 17,767 million tokens, so local hardware is mostly a fixed cost, not a per-token one. TDP is the peak draw, so this is an upper bound. Assumptions.
Run it
Pick your tool. All 2 load the same Q4_K_M weights.
llama-cli -hf lmstudio-community/sarvam-m-GGUF:Q4_K_M lms get lmstudio-community/sarvam-m-GGUF How to run it
On Windows use LM Studio (Best GUI on Windows, auto-detects CUDA/Vulkan backends.).
- Parameters
- 24B
- Q4_K_M size
- 14.3 GB
- Q8_0 size
- 25.1 GB
- Context
- 32k
- Memory
- 24 GB vram
- Usable for weights
- ~23 GB
- Power draw
- ~450 W
- Best runtime
- vLLM (Linux) / Ollama (CUDA)
You could also run
Run Sarvam-M 24B on other hardware
FAQ
Can Nvidia GeForce RTX 4090 (24GB) run Sarvam-M 24B?
Yes. Sarvam-M 24B runs on Nvidia GeForce RTX 4090 (24GB) at Q4_K_M (~16.3 GB of ~23 GB usable).
How much memory does Sarvam-M 24B need?
Nvidia GeForce RTX 4090 (24GB) has room to spare. At Q4_K_M the weights are ~14.3 GB; with KV cache and runtime overhead, budget ~16.3 GB at a 4k context.
What is the best tool to run Sarvam-M 24B on Windows?
LM Studio for a simple setup; Ollama (CUDA) for the most speed. AMD GPUs run via Vulkan/ROCm at roughly half CUDA throughput. NVIDIA is the smooth path on Windows.
Embed this
A live badge for your model card or README, updated as the data is. [](https://localmodel.run/can-i-run/sarvam-m-24b/nvidia-rtx-4090-24gb) Sources
Weights are measured from GGUF files; KV cache and overhead are computed, so totals can vary ~15% with context and runtime. Any tok/s is a bandwidth estimate. See methodology.