Amazon Bedrock
Try Amazon Bedrock on us. Starter credit covers your first requests, then one key reaches all 34 Amazon Bedrock models, with the privacy posture shown on every route.
Amazon Bedrockamazon-bedrock
Amazon Bedrock's regional OpenAI-compatible Mantle API. Credits routes use standard-tier pricing in US East (N. Virginia).
These privacy labels describe Amazon Bedrock, the upstream model provider. Confidential requires all three: verified provider compute, end-to-end encryption and explicit zero data retention. Account and billing metadata are outside the content-retention claim.
| Provider | Amazon Bedrock |
|---|---|
| API operator country | US Legal home of the API operator, not an inference-location guarantee. |
| Headquarters | Not verified |
| Routing status | Active |
| Provider website | https://aws.amazon.com/bedrock/ |
| Models | 34 public models |
| Credits routes | 34 |
| Zero data retention | not claimed |
| Verified confidential inference | Not verified |
| Policy note | Direct Amazon Bedrock Mantle inference through the US East (N. Virginia) regional endpoint. Only exact-priced, account-tested standard-tier models are enabled. No global or cross-region inference profiles are used. These routes are Standard: no contracted ZDR or confidential-inference guarantee is asserted. Model-specific data-retention terms apply. Policy source |
Documentation and policy sources
Inference locations
Provider API declarations are also available on each model's endpoint API. Direct chat and Responses replies separate advertised availability from serving location in usage.inference_location; the actual serving region remains unknown when not reported.
usage.provider_usage.gateway_region identifies the TrustedRouter gateway, not the provider's inference location. A value such as europe-west4 does not mean the model ran in Europe. usage.inference_location.serving_region remains null when the provider has not reported it. Older gateways may return the gateway-only field as provider_usage.region.
Where the upstream GPUs execute the model, not the provider's headquarters, API ingress, or data storage. ZDR and confidential inference do not by themselves restrict geography.
- Declared countries or regions
- Not verified
No inference-location declaration has been verified for this provider. - Can locations change?
- Not verified. Do not assume that the serving location is fixed.
- Provider region pinning
- Not verified.
- Pinning through TrustedRouter
- No verified provider-specific location pin documented here.
- Infrastructure declaration
- Inference location not verified; no country-specific residency commitment.
- Evidence
- Not yet reviewed.
Provider selection alone is not a region pin. TrustedRouter's US provider filter selects company jurisdiction, not GPU location. A regional gateway hostname does not pin a downstream GPU. For a binding residency requirement, confirm the model, region, failover scope and enforced controls before sending sensitive data.
Measured performance
30 samplesContinuously sampled across Amazon Bedrock's routed models: p50 TTFT, effective throughput, and success rate. Effective throughput uses provider-reported output tokens over complete request time. Unsupported route and probe-configuration rows are separated from provider downtime. No prompt or output content stored.
| p50 TTFT | 1348 ms |
|---|---|
| Effective throughput | 132 tok/s n=2 |
| Uptime | 100.00% |
| Model | p50 TTFT | Effective throughput | Uptime | Config excluded | Availability samples |
|---|---|---|---|---|---|
| qwen/qwen3-coder-next | 572 ms | — | 100.00% | — | 1 |
| deepseek/deepseek-v3.1 | 666 ms | — | 100.00% | — | 2 |
| mistralai/mistral-large-2512 | 833 ms | — | 100.00% | — | 1 |
| mistralai/magistral-small-2509 | 880 ms | — | 100.00% | — | 1 |
| qwen/qwen3-next-80b-a3b-instruct | 930 ms | — | 100.00% | — | 1 |
| openai/gpt-oss-safeguard-20b | 1031 ms | — | 100.00% | — | 2 |
| nvidia/nemotron-3-nano-30b-a3b | 1040 ms | — | 100.00% | — | 3 |
| google/gemma-3-27b-it | 1065 ms | — | 100.00% | — | 1 |
| minimax/minimax-m2 | 1256 ms | — | 100.00% | — | 2 |
| openai/gpt-oss-120b | 1348 ms | 132 tok/s n=2 | 100.00% | — | 2 |
| moonshotai/kimi-k2-thinking | 1393 ms | — | 100.00% | — | 1 |
| nvidia/nemotron-3-super-120b-a12b | 1539 ms | — | 100.00% | — | 1 |
| moonshotai/kimi-k2.5 | 1872 ms | — | 100.00% | — | 1 |
| mistralai/ministral-3-3b-instruct | 2115 ms | — | 100.00% | — | 4 |
| google/gemma-3-4b-it | 2346 ms | — | 100.00% | — | 1 |
| mistralai/ministral-3-8b-instruct | 2418 ms | — | 100.00% | — | 1 |
| google/gemma-3-12b-it | 2817 ms | — | 100.00% | — | 1 |
| qwen/qwen3-coder-30b-a3b-instruct | 2836 ms | — | 100.00% | — | 1 |
| z-ai/glm-4.7 | 2869 ms | — | 100.00% | — | 1 |
| minimax/minimax-m2.5 | 3350 ms | — | 100.00% | — | 1 |
| z-ai/glm-4.7-flash | 7452 ms | — | 100.00% | — | 1 |
Amazon Bedrock performance history · Full provider & model leaderboard.
Models served by Amazon Bedrock.
Each row links to pricing, provider, benchmark, and API pages for the model.
| Model | AI IQ | Context | Input | Cached input | Output |
|---|---|---|---|---|---|
deepseek/deepseek-v3.1deepseek-ai/DeepSeek-V3.1 |
IQ 98#131 | 163,840 | $0.6119/1M | Not published | $1.7724/1M |
deepseek/deepseek-v3.2DeepSeek: DeepSeek V3.2 |
IQ 102#110 | 131,072 | $0.6541/1M | Not published | $1.95175/1M |
google/gemma-3-12b-itGoogle: Gemma 3 12B |
— | 131,072 | $0.09495/1M | Not published | $0.30595/1M |
google/gemma-3-27b-itGoogle: Gemma 3 27B |
— | 131,072 | $0.24265/1M | Not published | $0.4009/1M |
google/gemma-3-4b-itGoogle: Gemma 3 4B |
— | 131,072 | $0.0422/1M | Not published | $0.0844/1M |
minimax/minimax-m2MiniMax: MiniMax M2 |
— | 204,800 | $0.3165/1M | Not published | $1.266/1M |
minimax/minimax-m2.1MiniMax: MiniMax M2.1 |
IQ 104#101 | 204,800 | $0.3165/1M | Not published | $1.266/1M |
minimax/minimax-m2.5MiniMax: MiniMax M2.5 |
IQ 104#102 | 196,608 | $0.3165/1M | Not published | $1.266/1M |
mistralai/devstral-2-123bdevstral-2-123b |
— | 4,096 | $0.422/1M | Not published | $2.11/1M |
mistralai/magistral-small-2509magistral-small-2509 |
— | 4,096 | $0.5275/1M | Not published | $1.5825/1M |
mistralai/ministral-3-14b-instructMinistral 3 14B Instruct |
— | 262,144 | $0.211/1M | Not published | $0.211/1M |
mistralai/ministral-3-3b-instructministral-3-3b-instruct |
— | 4,096 | $0.1055/1M | Not published | $0.1055/1M |
mistralai/ministral-3-8b-instructministral-3-8b-instruct |
— | 4,096 | $0.15825/1M | Not published | $0.15825/1M |
mistralai/mistral-large-2512Mistral: Mistral Large 3 2512 |
— | 262,144 | $0.5275/1M | Not published | $1.5825/1M |
moonshotai/kimi-k2-thinkingMoonshotAI: Kimi K2 Thinking |
IQ 92#152 | 262,144 | $0.633/1M | Not published | $2.6375/1M |
moonshotai/kimi-k2.5MoonshotAI: Kimi K2.5 |
IQ 110#77 | 262,144 | $0.633/1M | Not published | $3.165/1M |
nvidia/nemotron-3-nano-30b-a3bNVIDIA: Nemotron 3 Nano 30B A3B |
— | 262,144 | $0.0633/1M | Not published | $0.2532/1M |
nvidia/nemotron-3-super-120b-a12bNVIDIA: Nemotron 3 Super |
— | 262,144 | $0.15825/1M | Not published | $0.68575/1M |
nvidia/nemotron-nano-12b-v2nemotron-nano-12b-v2 |
— | 4,096 | $0.211/1M | Not published | $0.633/1M |
nvidia/nemotron-nano-9b-v2nemotron-nano-9b-v2 |
— | 4,096 | $0.0633/1M | Not published | $0.24265/1M |
openai/gpt-oss-120bOpenAI: gpt-oss-120b |
IQ 98#132 | 131,072 | $0.15825/1M | Not published | $0.633/1M |
openai/gpt-oss-20bOpenAI: gpt-oss-20b |
IQ 93#149 | 131,072 | $0.07385/1M | Not published | $0.3165/1M |
openai/gpt-oss-safeguard-120bGPT-OSS Safeguard 120B |
— | 131,072 | $0.15825/1M | Not published | $0.633/1M |
openai/gpt-oss-safeguard-20bgpt-oss-safeguard-20b |
— | 128,000 | $0.07385/1M | Not published | $0.211/1M |
qwen/qwen3-235b-a22b-instruct-2507Qwen/Qwen3-235B-A22B-Instruct-2507 |
— | 262,144 | $0.2321/1M | Not published | $0.9284/1M |
qwen/qwen3-32bQwen: Qwen3 32B |
— | 131,072 | $0.15825/1M | Not published | $0.633/1M |
qwen/qwen3-coder-30b-a3b-instructQwen: Qwen3 Coder 30B A3B Instruct |
— | 262,144 | $0.15825/1M | Not published | $0.633/1M |
qwen/qwen3-coder-480b-a35b-instructqwen3-coder-480b-a35b-instruct |
— | 256,000 | $0.47475/1M | Not published | $1.899/1M |
qwen/qwen3-coder-nextQwen: Qwen3 Coder Next |
— | 262,144 | $0.5275/1M | Not published | $1.266/1M |
qwen/qwen3-next-80b-a3b-instructQwen: Qwen3 Next 80B A3B Instruct |
— | 262,144 | $0.1477/1M | Not published | $1.266/1M |
qwen/qwen3-vl-235b-a22b-instructQwen: Qwen3 VL 235B A22B Instruct |
— | 131,072 | $0.55915/1M | Not published | $2.8063/1M |
writer/palmyra-vision-7bpalmyra-vision-7b |
— | 4,000 | $0.15825/1M | Not published | $0.633/1M |
z-ai/glm-4.7Z.ai: GLM 4.7 |
IQ 104#99 | 202,752 | $0.633/1M | Not published | $2.321/1M |
z-ai/glm-4.7-flashZ.ai: GLM 4.7 Flash |
— | 131,072 | $0.07385/1M | Not published | $0.422/1M |
Questions
Does Amazon Bedrock have zero data retention?
TrustedRouter does not currently mark Amazon Bedrock as provider-level zero data retention. Use trustedrouter/zdr or provider.min_privacy=zdr to select a different eligible route, and review the linked policy source for changes.
Is Amazon Bedrock end-to-end encrypted?
TrustedRouter does not currently mark Amazon Bedrock as end-to-end encrypted at the provider boundary. The TrustedRouter gateway is still attested, but the selected provider normally receives the request in order to run the model. Use trustedrouter/e2e for the stronger route requirement.
Which Amazon Bedrock models are available through TrustedRouter?
This page currently lists 34 public Amazon Bedrock models, with live pricing, route count, context length, measured performance when available, and links to each model's provider and benchmark pages.