I ran Gemma 4 and Qwen 3.5 for the same local tasks, and one pulled miles ahead
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
A hands-on comparison of two local LLMs — Google's Gemma 4 (E4B/E2B variants) and Qwen 3.5 9B — across everyday tasks like reasoning, document handling, and multimodal use. Qwen 3.5 9B wins on reasoning depth and long-context performance thanks to its hybrid GDN architecture and inherited thinking mode from Qwen 3. Gemma 4 wins overall for the author's workflow due to its faster latency, better image handling, lighter footprint, and mobile accessibility via PocketPal. The verdict: Qwen for research and heavy reasoning, Gemma for general daily use and visual tasks.
Table of contents
A quick overview of Qwen and GemmaJust talking to themEverything beyond typingWhere each one still has a place39 Impressions