xAI launches Grok 4.5 at roughly half the price of rival frontier models

xAI released Grok 4.5 — its first major launch since going public — priced at $2 per million input tokens and $6 per million output tokens. That compares with $5/$25 for Anthropic's Opus 4.8 ($10/$50 for its top-end Fable 5) and $5/$30 for OpenAI's GPT-5.6 Sol. Musk pitched it as "Opus-class at a fraction of the price," crediting a claimed 2x token efficiency and training data from coding tool Cursor. Independent evaluation by Artificial Analysis ranked it fourth on overall capability — genuinely competitive, not dominant.

Why it matters

The frontier price war just moved from the budget tier to the flagship tier. If an Opus-class-adjacent model at 50–80% off holds up in real workloads, it pressures exactly the margin OpenAI and Anthropic fund their compute buildouts with — the same squeeze Microsoft is applying internally with MAI, now applied from outside. The fourth-place independent ranking is the caveat that keeps this competitive rather than decisive.