Skip to main content
The Human Bit

xAI · Grok 4.5 · 16 July 2026

A new flagship is a reason to test, not a universal recommendation.

xAI launched Grok 4.5 for coding, agentic tasks and knowledge work, with configurable reasoning and access through Grok Build and the API.

WatchProvider announced
The Human Bit take

Watch, then test where coding or document production is already measurable. Do not migrate because one launch chart says ‘best.’

Why this may matter

A stronger option can improve a workflow or lower running cost. Provider benchmarks do not tell an owner whether the model understands their files, catches exceptions or reduces review time.

Keep human

Inspect every benchmark’s task, harness and comparison conditions. Then measure the job you actually pay people to do.

Inspect the change, example and test detail

What changed

The provider positions Grok 4.5 as its new flagship and default model in Grok Build, emphasising engineering work, office documents, speed and token efficiency.

Real example

A technical team should run its own bug-fix or spreadsheet task against the current tool, holding inputs constant and scoring correctness, elapsed time, cost and correction effort.

Test status

No test is attached to this item. Its current evidence state should not be read as a Human Bit result.

Sources and review

Sources retain their own provenance. A provider announcement establishes what was announced; it does not become independent testing.

Canonical claim: The provider positions Grok 4.5 as its new flagship and default model in Grok Build, emphasising engineering work, office documents, speed and token efficiency.

Readiness: Keep this visible without changing guidance until stronger evidence arrives.

Released
16 July 2026
Reviewed
25 July 2026
Review by
8 August 2026
Review owner
The Human Bit editorial
Revision
grok-45-launch@v1
Volatility
volatile