Simulating realistic Document summarization parameters (25,000 in / 600 out with 10% cache reuse). Qwen 3 32B Instruct delivers a 81% cost reduction over Amazon Nova Premier.
| Traffic Volume Tier | Qwen 3 32B Instruct Monthly | Amazon Nova Premier Monthly | Monthly Savings by picking Qwen 3 32B Instruct |
|---|---|---|---|
| 1,000 reqs/mo (Dev/Testing) | $9.58 | $50.30 | Save $40.72 / mo |
| 10,000 reqs/mo (Small App) | $95.80 | $503.00 | Save $407.20 / mo |
| 100,000 reqs/mo (Growth Production) | $958.00 | $5,030.00 | Save $4,072.00 / mo |
| 1,000,000 reqs/mo (Scale SaaS) | $9,580.00 | $50,300.00 | Save $40,720.00 / mo |
Qwen 3 32B Instruct is 81% cheaper for Document summarization workloads. At standard Document summarization parameter ratios (25,000 input tokens, 600 output tokens, 10% cache hit), Qwen 3 32B Instruct costs $0.00958 per request compared to $0.0503 on Amazon Nova Premier.
Qwen 3 32B Instruct offers a context window of 128,000 tokens (max output: 16,384), while Amazon Nova Premier offers 1,000,000 tokens (max output: 32,768).
At 100,000 requests per month, using Qwen 3 32B Instruct saves $4,072.00 every month (or $48,864.00 annually) compared to Amazon Nova Premier.
Long-context models pay off here — compare price per 1M tokens at your true document size. Summarize once, store the result; don't re-summarize unchanged documents. For batch backfills, nightly jobs can use cache-friendly request ordering.