Grok 4.3 becomes generally available on Amazon Bedrock

xAI's Grok 4.3 is now generally available on Amazon Bedrock, giving AWS customers direct, managed access to the model for building agents and AI workflows without leaving the Bedrock platform.

What's new

The model is available in Bedrock as xai.grok-4.3 and supports a 1 million token context window, configurable reasoning effort (none, low, medium, high), full function/tool calling through an OpenAI-compatible interface, structured JSON Schema output with strict-mode compliance, and multimodal input (text and images, text-out).

Access runs through Bedrock's Mantle inference engine via regional endpoints (for example https://bedrock-mantle.us-west-2.api.aws/openai/v1), authenticated with either short-term bearer tokens — recommended for production — or long-term API keys. AWS is offering three service tiers for the model: Standard, Priority, and Flex. Specific pricing is published on the Bedrock pricing page rather than in the announcement itself.

On the reasoning controls, AWS notes: "Drop the effort to none and that same field reports 0 reasoning tokens, which is the setting to reach for on simple, latency-sensitive calls." xAI describes the model's positioning as built "for enterprise work where accuracy matters," citing top rankings on multiple benchmarks and claiming "2 to 10 times more intelligence per dollar than other frontier models."

Context

Grok's arrival on Bedrock extends a pattern of frontier labs distributing their models through major cloud marketplaces rather than requiring direct API integration — the same route Anthropic's Claude and Meta's Llama models have used to reach AWS's enterprise customer base. For xAI, Bedrock availability lowers the integration barrier for enterprises already standardized on AWS infrastructure and procurement, sidestepping the need for a separate vendor relationship or billing setup with xAI directly.

Why it matters

Cloud-marketplace distribution is becoming a standard proof point for a frontier model's enterprise ambitions — being absent from Bedrock, Vertex AI, or Azure AI Foundry increasingly reads as a gap rather than a neutral choice. Grok's addition puts it alongside Anthropic, Meta, and Mistral models already on Bedrock, intensifying price and capability competition inside AWS's own catalog. The tiered service options (Standard/Priority/Flex) also signal AWS is treating Grok as a first-class citizen for latency- and cost-sensitive production workloads, not just an experimental add-on.