Grok 4.5 Places Behind Only Anthropic And OpenAI's Top Models On Artificial Analysis Intelligence Index

SpaceXAI's Grok 4.5 model placed fourth on the Artificial Analysis Intelligence Index with a score of 54, trailing only Anthropic's Claude Fable 5 and Claude Opus 4.8 and OpenAI's GPT-5.5, the benchmarking firm reported. The score represents a 16-point jump from Grok 4.3's score of 38 and puts xAI ahead of all open-weights models and Google's Gemini lineup, including Gemini 3.5 Flash and Gemini 3.1 Pro Preview. Artificial Analysis noted the scale of the jump is unusual for frontier model updates, which typically move by a handful of points. The improvement occurred while xAI managed co-founder departures and a public restructuring described by Elon Musk as deliberate. On the GDPval-AA v2 benchmark for real-world knowledge work, Grok 4.5 scored an Elo of 1543, placing between Claude Opus 4.8 at 1600 and GLM-5.2 at 1513, and ahead of GPT-5.5 at 1494. On the tau-cubed Banking test for agentic customer service, Grok 4.5 posted the highest score measured at 33%, edging GPT-5.5 at 31%. It also topped AutomationBench-AA, completing 51.4% of task objectives cleanly, ahead of Claude Fable 5 at 48.6% and Claude Opus 4.8 at 48.5%, while using roughly 8,000 output tokens per task, under a quarter of Opus 4.8's usage. Grok 4.5 runs at $0.31 per task on the Intelligence Index, roughly a fifth of Claude Sonnet 5 (max) cost, with listed pricing of $2 per million input tokens and $6 per million output tokens. Musk said Grok 4.5 is roughly three times the size of Grok 4.3 at 1.5 trillion parameters. Its hallucination rate climbed to 54% from 25%, and its context window shrank to 500,000 tokens from 1 million.