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

granite-4.2-3b

IBM
GermanApache 2.0

Open language model for chat and text. 3 sizes from 3.7B to 29.3B parameters.

Settings

Quality
4 bit (Q4_K_M) is the most common choice: good quality with low memory requirements.
Context length: 8k tokens
4k128k
How much text the model keeps in view at once. More context needs more memory.
Recommendation
1× H100 80 GB

The model needs about 4.4 GB of memory. Hopper, FP8 support.

Weights
2.2 GB
KV cache
0.7 GB
Speed
~400.3 Tok./s

More suitable configurations

CardQuantityTok./sStatus
Tesla T4 16 GB1×69.32-3 weeks delivery time
Tesla P100 16 GB1×87.32-3 weeks delivery time
Tesla P40 24 GB1×81.82-3 weeks delivery time
Tesla V100 16 GB1×192.9in stock
Tesla V100 32 GB1×192.9On order, delivery time approx. 1 week
AMD Instinct MI50 16 GB1×147.9in stock
AMD Instinct MI50 32 GB1×147.9in stock
AMD Instinct MI60 32 GB1×136.62-3 weeks delivery time
AMD Instinct MI100 32 GB1×165.12-3 weeks delivery time
A30 24 GB1×186.72-3 weeks delivery time
RTX A5000 24 GB1×194.52-3 weeks delivery time
A40 48 GB1×185.52-3 weeks delivery time