AxForge
Sign in Get started

Model reference · open weights

ThinkingCap-Qwen3.6

LLMs bottlecapai Vision + text 1 build Open weights 52k dl/mo

ThinkingCap-Qwen3.6 is an open-weight language model from bottlecapai. ThinkingCap-Qwen3.6-27B-NVFP4 (NVFP4) weighs 20.6 GB; the smallest configuration that runs it is RTX 4090 24 GB.

  • ThinkingCap-Qwen3.6 is an image-text-to-text model by bottlecapai that uses 16.7B parameters to reduce thinking tokens by 50% while preserving answer quality.
  • This specific release is an NVFP4 quantization that stores weights at 4 bits, reducing memory usage to approximately 19 GB and enabling faster decoding via vLLM.
  • The model is distributed under the apache-2.0 license.

Summary of the bottlecapai/ThinkingCap-Qwen3.6-27B-NVFP4 model card, 2026-10-01

Hugging Face ↗

What it is

Released bybottlecapai
Released2026-07-29
Parameters16.7B
VRAM20.6 GB for the weights

What it runs on

Memory and cards for ThinkingCap-Qwen3.6-27B-NVFP4 (NVFP4)

20.6 GBweights, file size
762 MBruntime overhead, at least

How much memory each request adds isn't estimated yet for this architecture. The weights need at least the cards below, plus room for the context.

CardWeights alone
RTX 3060 12 GB … RTX 4060 Ti 16 GB
2 smaller cards		does not fit
RTX 3090 24 GB
FP4 without its speed-up here		tight
RTX 4090 24 GB
FP4 without its speed-up here		tight
RTX 5090 32 GBfits
L40S 48 GB
FP4 without its speed-up here		fits
A100 80 GB
FP4 without its speed-up here		fits
H100 80 GB
FP4 without its speed-up here		fits
RTX PRO 6000 Blackwell 96 GBfits
DGX Spark (GB10) 128 GB unifiedfits
H200 141 GB
FP4 without its speed-up here		fits
B200 180 GBfits

How it works

How language models work

Your prompttext / messagesTransformerattention over tokensNext-token loopgenerate + streamResponsetext · tool callsA language model reads your tokens and predicts the next one, again and again, streaming the reply back.
© 2026 AxForge · EU-hosted AI infrastructure Pricing Docs Trust Privacy Terms