Qwen2.5-7B-Instruct vs granite-4.0-h-micro
Specs, VRAM requirements and download trends — updated 24 October 2026.
Qwen · A · Open weights
Params
7.6 B
Context
32,768
30d
6.4 M
apache-2.0 · commercial OK
ibm-granite · B · Open weights
Params
3.2 B
Context
131,072
30d
9 K
apache-2.0 · commercial OK
Specification comparison
Differences are highlighted; identical values are muted.
| Specification | Qwen2.5-7B-Instruct | granite-4.0-h-micro |
|---|---|---|
| Deployment | Open weights · also via API | Open weights · also via API |
| Parameters | 7.6 B | 3.2 B |
| Architecture | Qwen2ForCausalLM | GraniteMoeHybridForCausalLM |
| Context length | 32,768 | 131,072 |
| Licence | apache-2.0 | apache-2.0 |
| Commercial use | Allowed | Allowed |
| Languages | 1 | — |
| Downloads 30d | 6,369,723 | 9,011 |
| Downloads all time | 190.0 M | 254 K |
| Quantizations on the Hub | SAFETENSORS | SAFETENSORS |
| Gated | No | No |
| First seen on the Hub | 2024-09-16 | 2025-09-16 |
| API input price / 1M tokens | $0.10 | $0.017 |
| API output price / 1M tokens | $0.20 | $0.11 |
| Est. cost per 1M input + 1M output tokens | $0.30 | $0.13 |
| Cached input / 1M tokens | — | — |
| AA Intelligence Index | — | 5.1 |
| AA Coding Index | — | — |
| AA Agentic Index | — | — |
| Max output tokens | 29,491 | 117,900 |
| API providers | 1 | 1 |
| Tool calling | Yes | No |
| Reasoning mode | No | No |
API cost
Estimated cost of 1M input + 1M output tokens via OpenRouter
Qwen2.5-7B-Instruct
$0.30
granite-4.0-h-micro
$0.13
granite-4.0-h-micro is 2.3× cheaper at list price.
Download trend
Daily snapshots, last 89 days (28 Jul – 24 Oct)
12.3 M9 K
Qwen2.5-7B-Instruct
granite-4.0-h-micro
VRAM side by side
On RTX 4090 · 24 GB · 8K context unless noted
granite-4.0-h-micro · fp16 @ 128K ctx
15.2 GB / 24 GB
Qwen2.5-7B-Instruct · fp16
18.4 GB / 24 GB