Gemma 4 31B IT vs Qwen3.5-4B

Gemma 4 31B IT

A dense multimodal cousin with documented tool calling, optional thinking and a 256K context, for general agent work and image reading on one 24 GB GPU through Google's 4-bit build. It runs in every major runtime under plain Apache 2.0. The main risk is that Google publishes no safety results and has gone quiet on the repository since July.

Qwen3.5-4B

A small vision-language cousin for tool calls, light agent steps and image or video reading on almost any machine, with switchable thinking. Ollama's 3.3 GB build or the 9.3 GB BF16 files run in every major runtime under plain Apache 2.0. The main risks are no safety results, no structured-output guarantee and no answers from Alibaba since March.

Gemma 4 31B IT

1. On the reference machine run llama-server -hf google/gemma-4-31B-it-qat-q4_0-gguf --hf-file gemma-4-31B_q4_0-it.gguf --jinja -c 32768, adding gemma-4-31B-it-mmproj.gguf for images
2. Or ollama run gemma4:31b for the 20 GB Q4_K_M build
3. In vLLM serve with --enable-auto-tool-choice --tool-call-parser gemma4 --reasoning-parser gemma4, as vLLM's recipe does
4. Thinking is off by default, so pass enable_thinking: true in chat_template_kwargs to turn it on, and expect the reasoning in the thought channel block
5. Drop earlier thoughts from the history except on tool-call turns, as the card asks
6. Use the card's sampling, temperature 1.0, top_p 0.95, top_k 64
7. Validate function names and arguments before running a call, and check JSON output yourself or use the runtime's guided decoding
8. The w4a16 build relies on Marlin INT4 kernels that vLLM marks unsupported on AMD, so use the GGUF on the reference machine

Qwen3.5-4B

1. Smallest start is ollama run qwen3.5:4b (Q4_K_M, about 4 GB with context)
2. Full precision vllm serve Qwen/Qwen3.5-4B --port 8000 --max-model-len 262144 --reasoning-parser qwen3 --enable-auto-tool-choice --tool-call-parser qwen3_coder
3. Turn thinking off with "chat_template_kwargs": {"enable_thinking": false} in the request body, since it's on by default and /think and /nothink aren't supported
4. There's no generation_config.json, so pass the card's sampling yourself, for example temperature=0.7, top_p=0.8, top_k=20, presence_penalty=1.5 without thinking
5. Keep the context at 128K or more if you rely on thinking, and turn on YaRN only past 262,144 tokens
6. Validate any JSON it returns, since structured output isn't documented