Local LLM study6: Testing gemma4:e4b/qwen3.5:9b for document creation and summarization purposes
Unlike previous coding-related tasks, gemma4:e4b and qwen3.5:9b were tested for 'document creation and summarization'. A technical article of approximately 8,400 characters was provided with instructions to provide a Japanese summary in 3 to 5 bullet points.
While all previous verifications (study 1–5) were coding-related tasks, this study tested gemma4:e4b and qwen3.5:9b for the purpose of 'document creation and summarization'.
The verification method involved providing a technical article of approximately 8,400 characters and instructing, 'Please provide a Japanese summary in 3 to 5 bullet points.' Requests were sent directly to the Ollama API in a state with no background load.
As for the results, gemma4:e4b responded in one go in 93 seconds. In contrast, qwen3.5:9b required 178 seconds, but the content was 0 characters.
This item is a summary. The full, checkable version is at the original source:
Read the original at Zenn (LLM) ↗