Grok 4.6 Shines in a Test Against Grok Bot and Cursor Origin

xAI and Cursor have recently added Grok Bot, Grok 4.6, and Cursor Origin to their offerings. A benchmarking test ranked Grok 4.6 via the Claire Index, while prolonged usage highlighted multi-account connectors and a virtual machine in Grok Bot. Cursor Origin, on the other hand, was used as a replacement for GitHub to assess its actual capabilities.

Grok 4.6 evaluated and ranked via the Claire Index

Grok 4.6 was subjected to a benchmarking test called How I AI Vibe and was ranked on the Claire Index. The evaluation also compared Grok 4.6 to GPT-5.6 Sol, Claude Sonnet 5, and Opus 5. The results were presented in the form of a ranking on the index, with notable elements arising from design tests where Grok surprised the tester. One evaluation category was reported as unexpected compared to initial expectations.

Grok Bot: multi-account connectors and virtual machine

Five configurations of Grok Bot were set up to test the service. The multi-account connectors are highlighted as the key feature, complemented by an integrated virtual machine whose operation was detailed. The tool was used for a week in real-world conditions, with feedback on its limitations and concrete use cases accompanied by a verdict. The emphasis on this key feature motivated its use in the test, with some aspects described as unprecedented compared to other agent platforms. Despite these trials, OpenClaws continues to be used in parallel.

Cursor Origin tested as an alternative to GitHub

Cursor Origin was tested as an alternative to GitHub to evaluate its actual usage. Presented as a native replacement, the tool underwent detailed feedback. However, the tester has not adopted GitHub at this stage and specified the conditions that could convince them to switch. These trials take place in a context of closely timed launches, which also include Grok Bot and the Grok 4.6 model.