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

Qwen3-Coder-Next

Alibaba (Qwen)
Apache 2.0

Open model for programmers. 79.7B Parameter, partly mixture-of-experts (only a part of the parameters is active per token).

  • Mixture-of-experts: 79.7B parameters in total, of which 3.9B 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 49.5 GB of memory. Hopper, FP8 support.

Weights
47.8 GB
KV cache
0.2 GB
Speed
~419.1 Tok./s

More suitable configurations

CardQuantityTok./sStatus
Tesla T4 16 GB4×58.52-3 weeks delivery time
Tesla P100 16 GB4×73.72-3 weeks delivery time
Tesla P40 24 GB4×69.12-3 weeks delivery time
Tesla V100 16 GB4×162.9only 1 in stock
Tesla V100 32 GB2×187.0On order, delivery time approx. 1 week
AMD Instinct MI50 16 GB4×124.9only 2 in stock
AMD Instinct MI50 32 GB2×143.4in stock
AMD Instinct MI60 32 GB2×132.42-3 weeks delivery time
AMD Instinct MI100 32 GB2×160.12-3 weeks delivery time
A30 24 GB4×157.62-3 weeks delivery time
RTX A5000 24 GB4×164.22-3 weeks delivery time
A40 48 GB2×179.82-3 weeks delivery time