On June 28, 2026, Elon Musk said on X that xAI’s Grok 4.5, built on a 1.5‑trillion‑parameter V9 foundation model trained with Cursor coding data, has entered closed beta testing inside SpaceX and Tesla. Follow‑up coverage reports Musk claims Grok 4.5 approaches or may surpass Anthropic’s Claude Opus/Mythos family on internal benchmarks.
This article aggregates reporting from 4 news sources. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Grok 4.5 is xAI’s clearest attempt yet to play in the absolute frontier tier. A 1.5‑trillion‑parameter model trained with rich developer‑workflow data from Cursor suggests a deliberate push into deep coding and engineering assistance, not just generic chat. Running the model first inside Tesla and SpaceX gives xAI privileged access to massive, high‑value proprietary datasets—from vehicle telemetry to rocket operations—that can further sharpen performance and agentic capabilities.
In the AGI race, this matters because it shows Musk is doubling down on the “full‑stack” strategy: own the data‑generating systems (cars, rockets, satellites), own the compute (SpaceX’s Colossus infrastructure), and now own a frontier‑class model tightly integrated with both. If Grok 4.5 is even roughly competitive with GPT‑5.x and Claude Opus on reasoning and coding, it keeps xAI in the top cluster of labs rather than a second‑tier player.
The missing piece is transparency. So far, xAI has not released rigorous, independently replicable benchmarks or detailed safety documentation comparable to OpenAI or Anthropic. Until that happens, claims about surpassing Opus should be treated cautiously. But from a capabilities standpoint, the combination of V9 scale, Cursor data and in‑house deployment at two of the world’s most complex engineering companies is a signal that Grok is not standing still.



