SOURCE-REVIEWED MODEL PROFILE
Qwen3-4B-Instruct-2507
An instruction-tuned Qwen text model for conversation, writing, coding and multilingual tasks. This July 2025 Instruct release answers without a separate thinking mode.
Metadata reviewed 2026-10-24 · Original creator: Qwen
What it does
- Text conversation and instruction following
- Dense transformer architecture
- Non-thinking Instruct release
Local fit
CPU / small GPU. Try story dialogue or a turn-based adventure. Reasoning controls depend on backend support.
Native context reported by the card. Longer extensions need separate configuration and extra memory.
Evidence and limits
Source reviewed; no local generation benchmark for this exact checkpoint.
Connect a supported local text/vision backend; not independently tested
License and source
Keep upstream notices and check the exact downloaded artifact and parent-model terms. The original creator remains credited.
Read the original model card ↗
Reviewed source revision: cdbee75f17c01a7cc42f958dc650907174af0554