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

Qwen3.5-35B-A3B

Alibaba (Qwen)
Apache 2.0

Open language model for chat and text. 8 sizes from 900M to 403.4B parameters, partly mixture-of-experts (only a part of the parameters is active per token).

  • Mixture-of-experts: 36.0B parameters in total, of which 3.0B active per token. Memory requirement depends on all parameters, speed on the active ones.

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 23.3 GB of memory. Hopper, FP8 support.

Weights
21.6 GB
KV cache
0.2 GB
Speed
~543.0 Tok./s

More suitable configurations

CardQuantityTok./sStatus
Tesla T4 16 GB2×87.02-3 weeks delivery time
Tesla P100 16 GB2×109.62-3 weeks delivery time
Tesla P40 24 GB2×102.72-3 weeks delivery time
Tesla V100 16 GB2×242.3only 1 in stock
Tesla V100 32 GB1×261.7On order, delivery time approx. 1 week
AMD Instinct MI50 16 GB2×185.8in stock
AMD Instinct MI50 32 GB1×200.6in stock
AMD Instinct MI60 32 GB1×185.32-3 weeks delivery time
AMD Instinct MI100 32 GB1×224.02-3 weeks delivery time
A30 24 GB2×234.42-3 weeks delivery time
RTX A5000 24 GB2×244.32-3 weeks delivery time
A40 48 GB1×251.62-3 weeks delivery time