HS
HardwareStack

Can I run Mistral 7B v0.3 locally?

Yes. At Q4 with 8K context, Mistral 7B v0.3 needs about 6.1 GB, so it runs entirely on any GPU with 7 GB of VRAM or more, starting with the NVIDIA GeForce RTX 5060 Ti 8GB.

Memory needed by compression and context length

Compression4K context8K context32K context
Q4
The most common download. Small quality loss.
5.6 GB6.1 GB9.1 GB
Q5
Very close to full quality.
6.3 GB6.8 GB9.8 GB
Q8
Practically indistinguishable from FP16 at about half the size.
8.7 GB9.2 GB12 GB
FP16
Uncompressed. Reference quality, largest size.
15 GB16 GB19 GB

Total memory: model weights, context (KV cache) and about 1 GB of runtime overhead. To run entirely on the GPU, VRAM must be about 5% larger than this.

GPUs that run Mistral 7B v0.3 entirely on the GPU

  • NVIDIA GeForce RTX 5060 Ti 8GB

    8 GB · ~64 tok/s at Q4

    Amazon
  • NVIDIA GeForce RTX 5060

    8 GB · ~64 tok/s at Q4

    Amazon
  • AMD Radeon RX 9060 XT 8GB

    8 GB · ~47 tok/s at Q4

    Amazon
  • NVIDIA GeForce RTX 4060

    8 GB · ~41 tok/s at Q4

    Amazon
  • AMD Radeon RX 6650 XT

    8 GB · ~42 tok/s at Q4

    Amazon
  • AMD Radeon RX 5700 XT 50th Anniversary

    8 GB · ~64 tok/s at Q4

    Amazon
  • NVIDIA GeForce RTX 2060 SUPER

    8 GB · ~64 tok/s at Q4

    Amazon
  • AMD Radeon RX 7600

    8 GB · ~43 tok/s at Q4

    Amazon
  • AMD Radeon RX 6600 XT

    8 GB · ~38 tok/s at Q4

    Amazon
  • NVIDIA GeForce GTX 1080

    8 GB · ~47 tok/s at Q4

    Amazon

Model facts

Parameters
7.3B
Max context
32K tokens
License
Apache 2.0
Released
May 2024
View on Hugging Face

Check it with your hardware

Change the compression, context length and system RAM in the calculator.

Open calculator