xAI's Grok 4.3 cuts prices

xAI is betting that the next wave of model adoption comes from cheaper, fast-enough reasoning rather than another pure benchmark victory lap.

• –Official xAI docs now list `grok-4.3` as an API model with a 1M-token input context
• –Artificial Analysis reports a higher Intelligence Index score than Grok 4.20, plus a strong jump on real-world agentic tasks
• –Pricing appears to have come down sharply versus the prior generation, which matters more for production usage than a marginal leaderboard win
• –The positioning is pragmatic: keep Grok competitive on cost while still selling “truth-seeking” and reasoning as the differentiators
• –For developers, this is mainly interesting if xAI can sustain low-cost inference without trading away reliability