Run OpenClaw or Hermes on a local model: which fits your GPU

Ollama's default gemma4 is the small E4B and qwen3.5 is the 9B, not what most people assume. What each tag really downloads, what the bigger models…

Aliteq
Lena Fischer · AI & Local Compute Editor

The short answer

Run ollama launch openclaw or ollama launch hermes and pick a model, but check what the model name actually downloads. As of 25 Sep 2026, Ollama's default qwen3.5 tag is the 9B model (6.6GB), gemma4…

12GB card: Gemma 4 12B (≈8.8 GiB at 64K)

16GB card: gpt-oss 20B (≈15.6 GiB, tight) or the default gemma4 (E4B)

24GB card: Gemma 4 26B-A4B (≈17.2 GiB) or, tightly, Qwen3.6 35B-A3B (≈22.6 GiB)

Check the tag: qwen3.5 alone pulls the 9B, not the 35B

Aliteq

Read the full story

Run OpenClaw or Hermes on a local model: which fits your GPU

Read the full story on Aliteq