Qwen: Qwen2.5 VL 72B Instruct vs ScaleDown: Summarize
On input price, ScaleDown: Summarize is ~20.0× cheaper. Verified 2026-10-21.
Qwen: Qwen2.5 VL 72B Instruct $0.800/$1.00vsScaleDown: Summarize $0.040/Free
Verdict
On blended price (3:1 input:output), ScaleDown: Summarize is ~28.3× cheaper ($0.040/Free vs $0.800/$1.00 per 1M tokens).
Pricing
Per-million-token list prices, verified 2026-10-21. Green marks the better value.
| Metric | Qwen: Qwen2.5 VL 72B Instruct | ScaleDown: Summarize |
|---|---|---|
| Input / 1M | $0.800 | $0.040 |
| Output / 1M | $1.00 | Free |
| Cache read / 1M | $0.400 | — |
| Blended (3:1) | $0.850/1M | $0.030/1M |
| Estimated monthly cost | Qwen: Qwen2.5 VL 72B Instruct | ScaleDown: Summarize |
|---|---|---|
| Chatbot10,000 reqs · 800/400 tok | $10.40 | $0.32 |
| Coding agent5,000 reqs · 8000/2000 tok | $42.00 | $1.60 |
| RAG pipeline30,000 reqs · 4000/500 tok | $111.00 | $4.80 |
Plug in your own numbers in the calculator, or paste a real bill into the Bill Shredder.
Specifications & capabilities
| Metric | Qwen: Qwen2.5 VL 72B Instruct | ScaleDown: Summarize |
|---|---|---|
| Context window | 128K | 1M |
| Max output | 115K | 4K |
| Vision input | ✓ | ✓ |
| Tool calling | – | – |
| Reasoning | – | – |
Provider
Infrastructure metrics for the organizations behind these models.
| Metric | Qwen (Alibaba) | scaledown |
|---|---|---|
| Listed endpoints | 18 | — |
| Reliability | 99.3% | — |
| Avg latency | 450 ms | — |
| Pricing tiers | standard, free | — |
Full details: Qwen: Qwen2.5 VL 72B Instruct pricing · ScaleDown: Summarize pricing