Qwen 2.5 72B Instruct charges $0.35 per 1M input tokens and $0.70 per 1M output tokens (cached input: $0.035/M). Real bills depend on your input/output split — the tables below price Qwen 2.5 72B Instruct at 1K–1M tokens across common workload shapes. Verified Aug 19, 2026
Cost by token volume
| Volume | Input-heavy (90/10) | Balanced (70/30) | Output-heavy (30/70) | All input | All output |
|---|---|---|---|---|---|
| 1,000 tokens | $0.000385 | $0.000455 | $0.000595 | $0.00035 | $0.0007 |
| 10,000 tokens | $0.00385 | $0.00455 | $0.00595 | $0.0035 | $0.007 |
| 100,000 tokens | $0.0385 | $0.0455 | $0.0595 | $0.035 | $0.07 |
| 1 million tokens | $0.385 | $0.455 | $0.595 | $0.35 | $0.70 |
Uncached. Volume links open interactive pages with adjustable splits.
Prompt caching impact
| Cached share of input | Chat request (4K in / 800 out) | Monthly @ 10K req/day | Savings / month |
|---|---|---|---|
| 0% | $0.00196 | $588.00 | — |
| 50% | $0.00133 | $399.00 | −$189.00 |
| 75% | $0.001015 | $304.50 | −$283.50 |
Monthly projections
| Scale | Requests / day | Chat workload / month | Agentic workload / month* |
|---|---|---|---|
| 100 req/day | 3,000 | $3.99 | $20.58 |
| 1K req/day | 30,000 | $39.90 | $205.80 |
| 10K req/day | 300,000 | $399.00 | $2,058.00 |
| 100K req/day | 3,000,000 | $3,990.00 | $20,580.00 |
Chat: 4K in / 800 out, 50% cached. *Agentic: 40K in / 1.5K out, 65% cached. Model your exact mix in the monthly cost calculator.
FAQ
1M purely input tokens cost $0.35; 1M purely output tokens cost $0.70. A typical mixed workload lands in between — see the volume table for exact splits.
A chat-style request (4,000 input + 800 output) costs about $0.00133 with 50% cached input, or $0.00196 without caching.
At 1,000 requests/day of a chat workload (4K in / 800 out, 50% cached): $39.90/month. At 10,000 requests/day: $399.00/month. Model your exact mix with the monthly cost calculator.