← Evidence ledger
Capability gains

SpaceXAI releases Grok 4.5 for coding and agentic knowledge work

SpaceXAI released Grok 4.5 through its API, Grok Build, and Cursor, reporting stronger coding, engineering, and long-running agentic performance. The launch also priced broad developer access and positioned the exact version as the company's most capable model.

CURRENT ASSESSMENT · REVISION 1
TOWARD DOOM64confidence 83/100

Why it moved the index

The dated primary source reports a version-specific model with strong coding and engineering results, multi-hour agentic training rollouts, office-tool use, API access, and broad availability through Cursor and Grok Build. Benchmark evidence is largely developer-reported and the page provides limited independent safety evidence, so confidence is high but not maximal.

AUDIT TRAIL

Assessment history

  1. R1
    Toward 64 · confidence 83

    Historical gap fill from a dated primary launch source; no Grok 4.5 story or model existed in the durable context.

    11 Aug 2026