Elon Musk Claims Grok 4.9 Will Match Top Rivals and Grok 5 Could Achieve AGI
xAI CEO Elon Musk stated on social media that Grok 4.9 is targeted to match rival models like OpenAI Astra and Anthropic Fable, while positioning Grok 5 as the version that could achieve Artificial General Intelligence (AGI). He also mentioned that Grok 4.7 currently performs similarly to Anthropic Opus 5.0, though its multimodal capabilities still require fixes. These statements highlight xAI's aggressive strategy to keep pace with top frontier AI labs like OpenAI and Anthropic in the race toward general-purpose AI. However, because these projections rely heavily on social media claims without accompanying benchmark test results or technical papers, industry observers remain cautious. Musk revealed that the upcoming Grok 4.8 is a 2.5-trillion parameter model trained on a completely new C++ software stack. The company is actively addressing multimodal performance issues identified in Grok 4.7 before pushing ahead with larger iterations.
## BACKGROUND
Artificial General Intelligence (AGI) is a hypothetical level of machine intelligence where a system matches or exceeds human cognitive capabilities across virtually all intellectual tasks. Leading AI companies such as OpenAI (with GPT-6 Astra) and Anthropic (with Claude Fable) are continually pushing frontier models toward long-horizon reasoning and autonomous task execution.