• 3 % discount on prepayment
  • Insured shipping from Switzerland
  • Sustainable: refurbished instead of new
  • Warranty on all cards
← Back to overview

Devstral-Small-2-24B-Instruct-2512

Mistral AI
Apache 2.0

Open model for programmers. 2 sizes from 24B to 125B parameters.

Settings

Quality
4 bit (Q4_K_M) is the most common choice: good quality with low memory requirements.
Context length: 8k tokens
4k256k
How much text the model keeps in view at once. More context needs more memory.
Recommendation
1× H100 80 GB

The model needs about 17.2 GB of memory. Hopper, FP8 support.

Weights
14.4 GB
KV cache
1.3 GB
Speed
~67.9 Tok./s

More suitable configurations

CardQuantityTok./sStatus
Tesla T4 16 GB2×10.92-3 weeks delivery time
Tesla P100 16 GB2×13.72-3 weeks delivery time
Tesla P40 24 GB1×13.92-3 weeks delivery time
Tesla V100 16 GB2×30.3only 1 in stock
Tesla V100 32 GB1×32.7On order, delivery time approx. 1 week
AMD Instinct MI50 16 GB2×23.2in stock
AMD Instinct MI50 32 GB1×25.1in stock
AMD Instinct MI60 32 GB1×23.22-3 weeks delivery time
AMD Instinct MI100 32 GB1×28.02-3 weeks delivery time
A30 24 GB1×31.72-3 weeks delivery time
RTX A5000 24 GB1×33.02-3 weeks delivery time
A40 48 GB1×31.52-3 weeks delivery time