xAI's Composer 2.5 & Grok Build 0.1 boost AI coding speed

xAI has rolled out a series of updates to its Grok ecosystem, headlined by the Composer 2.5 model now available to SuperGrok and X Premium+ users. Composer 2.5 is a fast, state-of-the-art model optimized for long-running tasks and following complex instructions, accessible via the /models menu in Grok Build. Alongside this, xAI launched Grok Build 0.1 in public beta on the xAI API—a coding model trained specifically for agentic coding tasks like web development, debugging, and MCP support. The model serves at over 100 tokens per second and is priced at $1 per million input tokens and $2 per million output tokens, making it a speedy, economical option for general-purpose agentic and tool-calling use cases.

Further enhancements include Smart Turn end-of-turn detection for the streaming Speech-to-Text API, which uses a machine learning model to predict when a speaker has finished their thought, reducing false endpoints during dictation. The Context Compaction API shrinks long conversations into shorter contexts, lowering costs and improving response times for agent loops. Additionally, the WebSocket Responses API mode provides a single, long-lived connection for lower end-to-end latency on tool-heavy workloads. These updates collectively position Grok as a more powerful platform for developers building agentic applications, offering faster performance, reduced costs, and improved accuracy.