Amazon Bedrock vs OpenAI. Compare a regional Bedrock quote with an OpenAI direct offering in a nearby workload-cost band.
USD token-only estimates using each offering’s published default tier. 1M tokens are an aggregate workload, not a single request. Cache savings, batch discounts, long-context surcharges, tools, taxes and negotiated rates are excluded. Tokenizers and capabilities may differ; lower price does not establish equivalent quality.
Amazon Bedrock · Amazon Nova Pro · Bedrock us-east-1 is 71.4% cheaper for this workload. Absolute difference: $10.00.
| Metric | Amazon Nova Pro · Bedrock us-east-1Amazon Bedrock | GPT-5.6 TerraOpenAI |
|---|---|---|
| Input / 1M tokens | $0.80 | $2.00 |
| Output / 1M tokens | $3.20 | $12.00 |
| Cached input / 1M | $0.20 | $0.20 |
| Context window | Not published | 1,050,000 |
| Maximum output | Not published | 128,000 |
| 1M input + 1M output | $4.00 | $14.00 |
| Default pricing basis | standard | standard |
| Last verified (UTC) | 2026-09-06T12:40:24.070Z | 2026-09-04T18:24:25.246Z |
Each calculator uses the same default-tier assumption described above.
Amazon Bedrock On-Demand Inference token rates for US East (N. Virginia), not an Amazon Nova direct API global price. Cache-write, batch, priority, flex, latency-optimized, cross-region and non-text charges are excluded.
Official pricing sourceLast verified:
All tiers and model details →See the official pricing source for billing conditions.
Official pricing sourceLast verified:
All tiers and model details →