Grok 4.3 (low) is a reasoning-enabled configuration of xAI's flagship large language model, released in May 2026. Operating at the "low" reasoning effort level by default, this model is optimized for high-throughput agentic workflows that require advanced logic and instruction-following without the latency of high-intensity reasoning passes. It represents a significant iteration in the Grok 4 series, emphasizing improved factual reliability and reduced hallucination rates compared to earlier versions.
The model features a 1 million token context window, allowing it to ingest and analyze massive datasets, including extensive code repositories and long-form technical documentation. As a multimodal system, Grok 4.3 (low) processes both text and image inputs, enabling complex visual reasoning tasks such as diagram interpretation and multi-source data synthesis. It is natively integrated with web and X search tools, providing the ability to reference real-time information that extends beyond its December 2025 knowledge cutoff.
Grok 4.3 (low) is designed for agent-based and productivity-focused workflows, demonstrating strong performance on benchmarks such as GDPval-AA and the Artificial Analysis Intelligence Index. Key capabilities include parallel function calling, structured outputs, and native file generation for assets like PDFs and spreadsheets. With a scale of approximately 500 billion parameters, the model is positioned as a cost-efficient alternative for developers building complex, multi-step automation pipelines.