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

Qwen3.6-27B

Alibaba (Qwen)
Apache 2.0

Open language model for chat and text. 2 sizes from 27.8B to 36B parameters, partly mixture-of-experts (only a part of the parameters is active per token).

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 18.7 GB of memory. Hopper, FP8 support.

Weights
16.7 GB
KV cache
0.5 GB
Speed
~60.4 Tok./s

More suitable configurations

CardQuantityTok./sStatus
Tesla T4 16 GB2×9.72-3 weeks delivery time
Tesla P100 16 GB2×12.22-3 weeks delivery time
Tesla P40 24 GB1×12.32-3 weeks delivery time
Tesla V100 16 GB2×26.9only 1 in stock
Tesla V100 32 GB1×29.1On order, delivery time approx. 1 week
AMD Instinct MI50 16 GB2×20.7in stock
AMD Instinct MI50 32 GB1×22.3in stock
AMD Instinct MI60 32 GB1×20.62-3 weeks delivery time
AMD Instinct MI100 32 GB1×24.92-3 weeks delivery time
A30 24 GB1×28.12-3 weeks delivery time
RTX A5000 24 GB1×29.32-3 weeks delivery time
A40 48 GB1×28.02-3 weeks delivery time