Qwen3-4B-Instruct-2507 vs TwIL-LM3

Specs, VRAM requirements and download trends — updated 21 September 2026.

Qwen · A
Params
4.0 B
Context
262,144
30d
4.0 M
apache-2.0 · commercial OK
webAI-Official · B
Params
3.1 B
Context
65,536
30d
384 K

Specification comparison

Differences are highlighted; identical values are muted.

Specification Qwen3-4B-Instruct-2507 TwIL-LM3
Parameters 4.0 B 3.1 B
Architecture Qwen3ForCausalLM SmolLM3ForCausalLM
Context length 262,144 65,536
Licence apache-2.0 other
Commercial use Allowed Unknown
Languages 1
Downloads 30d 3,978,519 383,511
Downloads all time 61.8 M 495 K
Quantizations on the Hub SAFETENSORS GGUF, SAFETENSORS
Gated No No
First seen on the Hub 2025-08-05 2026-08-03

Download trend

Daily snapshots, last 56 days (28 Jul – 21 Sep)

4.0 M179 K
Qwen3-4B-Instruct-2507 TwIL-LM3

VRAM side by side

On RTX 4090 · 24 GB · 8K context unless noted

TwIL-LM3 · Q4_K_M 3.1 GB / 24 GB
✅ Runs comfortably
TwIL-LM3 · fp16 7.7 GB / 24 GB
✅ Runs comfortably
Qwen3-4B-Instruct-2507 · fp16 @ 256K ctx 28.6 GB / 24 GB
❌ Won’t fit

Adjacent comparisons