Qwen2.5-7B-Instruct vs Qwen3-4B-Instruct-2507-unsloth-bnb-4bit
Specs, VRAM requirements and download trends — updated 25 October 2026.
Qwen · A · Open weights
Params
7.6 B
Context
32,768
30d
6.1 M
apache-2.0 · commercial OK
unsloth · B · Open weights
Params
4.0 B
Context
262,144
30d
153 K
apache-2.0 · commercial OK
Specification comparison
Differences are highlighted; identical values are muted.
| Specification | Qwen2.5-7B-Instruct | Qwen3-4B-Instruct-2507-unsloth-bnb-4bit |
|---|---|---|
| Deployment | Open weights · also via API | Open weights |
| Parameters | 7.6 B | 4.0 B |
| Architecture | Qwen2ForCausalLM | Qwen3ForCausalLM |
| Context length | 32,768 | 262,144 |
| Licence | apache-2.0 | apache-2.0 |
| Commercial use | Allowed | Allowed |
| Languages | 1 | — |
| Downloads 30d | 6,077,597 | 152,773 |
| Downloads all time | 190.2 M | 1.8 M |
| Quantizations on the Hub | SAFETENSORS | SAFETENSORS |
| Gated | No | No |
| First seen on the Hub | 2024-09-16 | 2025-08-06 |
| API input price / 1M tokens | $0.10 | not on OpenRouter |
| API output price / 1M tokens | $0.20 | not on OpenRouter |
| Est. cost per 1M input + 1M output tokens | $0.30 | not on OpenRouter |
| Cached input / 1M tokens | — | not on OpenRouter |
| AA Intelligence Index | — | not on OpenRouter |
| AA Coding Index | — | not on OpenRouter |
| AA Agentic Index | — | not on OpenRouter |
| Max output tokens | 29,491 | not on OpenRouter |
| API providers | 1 | not on OpenRouter |
| Tool calling | Yes | not on OpenRouter |
| Reasoning mode | No | not on OpenRouter |
API cost
Estimated cost of 1M input + 1M output tokens via OpenRouter
Qwen2.5-7B-Instruct
$0.30
Download trend
Daily snapshots, last 90 days (28 Jul – 25 Oct)
12.3 M150 K
Qwen2.5-7B-Instruct
Qwen3-4B-Instruct-2507-unsloth-bnb-4bit
VRAM side by side
On RTX 4090 · 24 GB · 8K context unless noted
Qwen2.5-7B-Instruct · fp16
18.4 GB / 24 GB
Qwen3-4B-Instruct-2507-unsloth-bnb-4bit · fp16 @ 256K ctx
23.7 GB / 24 GB