Qwen3.8-27B vs Qwen2.5-VL-72B-Instruct
Specs, VRAM requirements and download trends — updated 24 October 2026.
Specification comparison
Differences are highlighted; identical values are muted.
| Specification | Qwen3.8-27B | Qwen2.5-VL-72B-Instruct |
|---|---|---|
| Deployment | Open weights · also via API | Open weights · also via API |
| Parameters | 27.8 B | 73.4 B |
| Architecture | Qwen3_5ForConditionalGeneration | Qwen2_5_VLForConditionalGeneration |
| Context length | 262,144 | 128,000 |
| Licence | apache-2.0 | other |
| Commercial use | Allowed | Unknown |
| Languages | — | 1 |
| Downloads 30d | 7,710,011 | 89,270 |
| Downloads all time | 17.7 M | 7.7 M |
| Quantizations on the Hub | SAFETENSORS | SAFETENSORS |
| Gated | No | No |
| First seen on the Hub | 2026-08-05 | 2025-01-27 |
| API input price / 1M tokens | $0.20 | $0.80 |
| API output price / 1M tokens | $2.50 | $1.00 |
| Est. cost per 1M input + 1M output tokens | $2.70 | $1.80 |
| Cached input / 1M tokens | $0.09 | $0.40 |
| AA Intelligence Index | 33.7 | — |
| AA Coding Index | 68.1 | — |
| AA Agentic Index | 45.8 | — |
| Max output tokens | 235,929 | 115,200 |
| API providers | 21 | 1 |
| Tool calling | Yes | No |
| Reasoning mode | Yes | No |
API cost
Estimated cost of 1M input + 1M output tokens via OpenRouter
Qwen3.8-27B
$2.70
Qwen2.5-VL-72B-Instruct
$1.80
Qwen2.5-VL-72B-Instruct is 1.5× cheaper at list price.
Download trend
Daily snapshots, last 66 days (20 Aug – 24 Oct)
7.8 M59 K
Qwen3.8-27B
Qwen2.5-VL-72B-Instruct
VRAM side by side
On RTX 4090 · 24 GB · 8K context unless noted
Qwen3.8-27B · fp16
65.8 GB / 24 GB
Qwen2.5-VL-72B-Instruct · fp16 @ 125K ctx
334.1 GB / 24 GB