Summary
xAI announced Grok 4.5 on July 8, 2026 and opened public access on July 9, positioning it as its first model built specifically for coding and agentic work. Built on a 1.5-trillion-parameter foundation and trained partly on real Cursor developer session data, it ships with configurable reasoning effort and is priced at 2 dollars per million input tokens and 6 dollars per million output tokens.
What changed
xAI released Grok 4.5 for coding and agentic tasks with configurable reasoning effort and per-token pricing of 2 dollars input and 6 dollars output per million tokens, following a July 8 announcement and July 9 public availability.
Why it matters
Grok 4.5 is xAI's explicit pivot toward developers, and its training on real Cursor session data shows how coding-tool usage becomes a data moat for model quality. At 2-and-6 pricing it undercuts flagship frontier models on cost, intensifying the price-performance competition for agentic coding work.
Evidence excerpt
Grok 4.5, released July 8-9, 2026, is xAI's first model built for coding and agentic work, trained on real Cursor session data, on a 1.5T foundation, priced at 2 dollars input / 6 dollars output per million tokens with configurable reasoning effort.