xAI’s Grok 4.3 is now available on Amazon Bedrock, bringing a one million token context window, a forty percent reduction in input pricing, and a sixty percent reduction in output pricing compared to the previous Grok 4 model. Enterprise developers using Amazon’s cloud infrastructure can access the model directly without going through xAI’s standalone API.

A one million token context window matters in practical terms. It means the model can hold and process roughly 750,000 words in a single interaction — enough for an entire legal document corpus, a full technical codebase, or several years of business communications as part of one query. Use cases that have been limited by context window size — extended document analysis, long-form code review, multi-session conversation history — become significantly more workable at this scale.

The pricing reduction is also meaningful at scale. AI inference costs have been falling across major model providers through 2026 as competition has intensified, but a forty percent input reduction from one version to the next sits toward the larger end of what the market has seen. For companies running high-volume AI workflows, the difference between Grok 4 and Grok 4.3 pricing could amount to substantial savings at enterprise usage levels.

xAI has highlighted improved agentic performance in Grok 4.3, meaning the model handles multi-step tasks, tool use, and sequential decision-making more reliably than its predecessor. Agentic AI — systems that complete complex tasks without human oversight at each step — has been one of the central competitive battlegrounds among AI model providers this year, and better agentic performance is one of the features enterprise customers consistently request.

Amazon Bedrock is AWS’s managed AI model marketplace, offering enterprise customers access to models from multiple providers — including Anthropic’s Claude, Meta’s Llama, and now xAI’s Grok — through a single API with consistent authentication, billing, and compliance infrastructure. Adding Grok 4.3 to that catalogue extends xAI’s reach into the enterprise market beyond what its standalone API has covered so far.

Benchmark performance data across standard evaluation datasets has not yet been published, though xAI’s documentation confirms the context window and pricing figures.

Grok 4.3 is available now on Amazon Bedrock. More on AI model releases and the enterprise AI market in 2026 is in our tech section. Our coverage of the competitive AI model pricing landscape tracks changes across major providers. The xAI product history and Grok development timeline is in our AI archives.