API compatibility - Amazon Bedrock
AI21AmazonAnthropicCohereDeepSeekGoogleMetaMiniMaxMistralMoonshotNVIDIAOpenAIQwenStabilityTwelveLabsWriterZ.AIModels supporting StartAsyncInvokeInvokeModelWithBidirectionalStream

API compatibility

Amazon Bedrock supports three families of runtime APIs, each designed for different integration patterns and use cases.

Invoke family: InvokeModel handles synchronous, single-response calls. InvokeModelWithResponseStream returns responses as a real-time stream. InvokeModelWithBidirectionalStream enables full-duplex streaming for interactive applications. AsyncInvoke submits long-running requests asynchronously, storing output to Amazon S3.

Converse family: Converse provides a unified, model-agnostic interface for synchronous multi-turn conversations. ConverseStream delivers the same experience with streaming output.

OpenAI-compatible family: ChatCompletions implements the OpenAI Chat Completions interface, enabling existing OpenAI-based integrations to run on Bedrock with minimal changes. Responses API implements the OpenAI Responses interface, supporting stateful, agentic interactions with built-in tool use and conversation history management.

We will now look at the list of APIs supported by each model.

AI21

Model name Invoke Converse Chat Completions Responses
Jamba 1.5 Large* Yes Yes No No
Jamba 1.5 Mini* Yes Yes No No

Amazon

Model name Invoke Converse Chat Completions Responses
Amazon Nova Multimodal Embeddings Yes No No No
Nova 2 Lite* Yes Yes No No
Nova 2 Sonic No No No No
Nova Canvas Yes No No No
Nova Lite* Yes Yes No No
Nova Micro* Yes Yes No No
Nova Premier* Yes Yes No No
Nova Pro* Yes Yes No No
Nova Reel No No No No
Nova Sonic* Yes No No No
Titan Embeddings G1 - Text Yes No No No
Titan Image Generator G1 v2 Yes No No No
Titan Multimodal Embeddings G1 Yes No No No
Titan Text Embeddings V2 Yes No No No
Titan Text Large No Yes No No

Anthropic

Model name Invoke Converse Chat Completions Responses
Claude 3 Haiku* Yes Yes No No
Claude 3.5 Haiku* Yes Yes No No
Claude Haiku 4.5* Yes Yes No No
Claude Opus 4.1* Yes Yes No No
Claude Opus 4.5* Yes Yes No No
Claude Opus 4.6* Yes Yes No No
Claude Sonnet 4* Yes Yes No No
Claude Sonnet 4.5* Yes Yes No No
Claude Sonnet 4.6* Yes Yes No No

Cohere

Model name Invoke Converse Chat Completions Responses
Command R* Yes Yes No No
Command R+* Yes Yes No No
Embed English Yes No No No
Embed Multilingual Yes No No No
Embed v4 Yes No No No
Rerank 3.5 Yes No No No

DeepSeek

Model name Invoke Converse Chat Completions Responses
DeepSeek V3.2* Yes Yes Yes No
DeepSeek-R1* Yes Yes No No
DeepSeek-V3.1* Yes Yes Yes No

Google

Model name Invoke Converse Chat Completions Responses
Gemma 3 12B IT* Yes Yes Yes No
Gemma 3 27B PT* Yes Yes Yes No
Gemma 3 4B IT* Yes Yes Yes No

Meta

Model name Invoke Converse Chat Completions Responses
Llama 3 70B Instruct* Yes Yes No No
Llama 3 8B Instruct* Yes Yes No No
Llama 3.1 405B Instruct No Yes No No
Llama 3.1 70B Instruct* Yes Yes No No
Llama 3.1 8B Instruct* Yes Yes No No
Llama 3.2 11B Instruct* Yes Yes No No
Llama 3.2 1B Instruct* Yes Yes No No
Llama 3.2 3B Instruct* Yes Yes No No
Llama 3.2 90B Instruct* Yes Yes No No
Llama 3.3 70B Instruct* Yes Yes No No
Llama 4 Maverick 17B Instruct* Yes Yes No No
Llama 4 Scout 17B Instruct* Yes Yes No No

MiniMax

Model name Invoke Converse Chat Completions Responses
MiniMax M2* Yes Yes Yes No
MiniMax M2.1* Yes Yes Yes No

Mistral

Model name Invoke Converse Chat Completions Responses
Devstral 2 123B* Yes Yes Yes No
Magistral Small 2509* Yes Yes Yes No
Ministral 14B 3.0* Yes Yes Yes No
Ministral 3 8B* Yes Yes Yes No
Ministral 3B* Yes Yes Yes No
Mistral 7B Instruct* Yes Yes No No
Mistral Large* Yes Yes No No
Mistral Large 3* Yes Yes Yes No
Mistral Small* Yes Yes No No
Mixtral 8x7B Instruct* Yes Yes No No
Pixtral Large* Yes Yes No No
Voxtral Mini 3B 2507* Yes Yes Yes No
Voxtral Small 24B 2507* Yes Yes Yes No

Moonshot

Model name Invoke Converse Chat Completions Responses
Kimi K2 Thinking* Yes Yes No No
Kimi K2.5* Yes Yes Yes No

NVIDIA

Model name Invoke Converse Chat Completions Responses
NVIDIA Nemotron Nano 12B v2 VL BF16* Yes Yes Yes No
NVIDIA Nemotron Nano 9B v2* Yes Yes Yes No
Nemotron Nano 3 30B* Yes Yes Yes No

OpenAI

Model name Invoke Converse Chat Completions Responses
GPT OSS Safeguard 120B* Yes Yes Yes Yes
GPT OSS Safeguard 20B* Yes Yes Yes Yes
gpt-oss-120b* Yes Yes Yes Yes
gpt-oss-20b* Yes Yes Yes Yes

Qwen

Model name Invoke Converse Chat Completions Responses
Qwen3 235B A22B 2507* Yes Yes Yes No
Qwen3 32B* Yes Yes Yes No
Qwen3 Coder 480B A35B Instruct* Yes Yes Yes No
Qwen3 Coder Next* Yes Yes Yes No
Qwen3 Next 80B A3B* Yes Yes Yes No
Qwen3 VL 235B A22B* Yes Yes Yes No
Qwen3-Coder-30B-A3B-Instruct* Yes Yes Yes No

Stability

Model name Invoke Converse Chat Completions Responses
Stable Image Conservative Upscale Yes No No No
Stable Image Control Sketch Yes No No No
Stable Image Control Structure Yes No No No
Stable Image Creative Upscale Yes No No No
Stable Image Erase Object Yes No No No
Stable Image Fast Upscale Yes No No No
Stable Image Inpaint Yes No No No
Stable Image Outpaint Yes No No No
Stable Image Remove Background Yes No No No
Stable Image Search and Recolor Yes No No No
Stable Image Search and Replace Yes No No No
Stable Image Style Guide Yes No No No
Stable Image Style Transfer Yes No No No

TwelveLabs

Model name Invoke Converse Chat Completions Responses
Marengo Embed 3.0 Yes No No No
Marengo Embed v2.7 No No No No
Pegasus v1.2 Yes No No No

Writer

Model name Invoke Converse Chat Completions Responses
Palmyra X4* Yes Yes No No
Palmyra X5* Yes Yes No No

Z.AI

Model name Invoke Converse Chat Completions Responses
GLM 4.7* Yes Yes Yes No
GLM 4.7 Flash* Yes Yes Yes No
Note

* Streaming Support: Models marked with an asterisk (*) also support InvokeModelWithResponseStream, which returns responses as a real-time stream.

Models supporting StartAsyncInvoke

StartAsyncInvoke is an Amazon Bedrock Runtime API that allows callers to submit a model invocation request and immediately receive back an invocationArn without waiting for the model to finish processing. The job runs in the background, and the output is written to a caller-specified S3 bucket once complete. Callers can then poll job status using the companion GetAsyncInvoke and ListAsyncInvokes APIs. The pattern is purpose-built for workloads involving large or latency-insensitive inputs, particularly video, audio, and bulk embedding generation, where holding an open synchronous connection would be impractical.

In terms of which models support it, the following models support StartAsyncInvoke:

  • TwelveLabs Marengo Embed 2.7 (twelvelabs.marengo-embed-2-7-v1:0) — required for video and audio input; InvokeModel only handles text and image

  • TwelveLabs Marengo Embed 3.0 (twelvelabs.marengo-embed-3-0-v1:0) — same pattern; async required for video/audio at scale

  • Amazon Nova Reel (amazon.nova-reel-v1:0 and v1:1) — video generation is exclusively async; output lands in S3

  • Amazon Nova Multimodal Embeddings (amazon.nova-2-multimodal-embeddings-v1:0) — async is required for video inputs larger than 25MB base64-encoded; sync is available for text, image, and document inputs

InvokeModelWithBidirectionalStream

InvokeModelWithBidirectionalStream is an Amazon Bedrock Runtime API that establishes a persistent, full-duplex channel between the caller and the model, allowing audio data to flow in both directions simultaneously and continuously. Unlike the standard InvokeModel or even InvokeModelWithResponseStream APIs, which follow a request-then-response pattern, this API keeps the connection open for the duration of a session so that the model can process incoming audio as it arrives and stream generated speech back in near real-time, without waiting for a complete utterance to finish. The interaction is structured around three phases: session initialization (where the client sends configuration events to set up the stream), audio streaming (where captured audio is encoded and sent as a continuous event stream), and response streaming (where the model simultaneously returns text transcriptions of user speech and synthesized audio output). InvokeModelWithBidirectionalStream cannot be used with Amazon Bedrock API keys and requires standard AWS credential-based authentication, reflecting its more complex session lifecycle compared to other Bedrock Runtime operations.

The following models support this API:

  • Amazon Nova Sonic family: Both amazon.nova-sonic-v1:0 and amazon.nova-2-sonic-v1:0 use it as their sole invocation path, since the speech-to-speech architecture fundamentally requires a live bidirectional channel that neither InvokeModel nor Converse can provide.