The Vibe Father
Grok 4.6 does not have a confirmed public launch record as of August 8, 2026. Elon Musk described its timing as “around August 7,” which is a target—not the same thing as a model card, API identifier, changelog, pricing page, or independently checkable benchmark result.
We checked xAI’s public model documentation on August 8. Its current guidance still tells developers to use Grok 4.5 for general work including code. That document is not a release announcement for 4.6, and it does not list a Grok 4.6 identifier, pricing, or technical profile. Until xAI publishes those details, the responsible status is unconfirmed.
What Musk actually signalled
Reporting on Musk’s post says Grok 4.6 was planned as a 1.5-trillion-parameter model with improved supervised fine-tuning and reinforcement learning. Grok 4.7 was described as a larger follow-up. Those are product claims and roadmap signals, not an independently measured comparison. They help explain what xAI intends to ship, they do not establish that it shipped on a particular day.
Why VibeBench is not assigning a score
A release date cannot substitute for evidence. A useful benchmark entry needs a stable model identity, a public way to access or inspect the release, and comparable evidence for its intelligence, coding, and agentic behavior. Vendor launch charts can be useful context when their methods are clear, but they still need labels and source links. A predicted score, a parameter count, or a social post does not belong in a ranked leaderboard.
| Known now | Still needed |
|---|---|
| Musk publicly indicated an around-August-7 target. | An xAI release announcement or public API model ID. |
| xAI’s public docs currently recommend Grok 4.5 for coding. | Pricing, context limits, availability, and a model card. |
| Grok 4.6 was described as a post-training-focused update. | Comparable, source-backed benchmark inputs. |
What changes when it arrives
When xAI publishes Grok 4.6, we will first verify the identifier and product details against xAI’s own materials. We will then add launch-period claims separately from the Vibe Coding Index and wait for compatible external sources before ranking it. That ordering matters, it lets readers see the launch news without confusing marketing, prediction, and measurement.
Bottom line
August 7 passed without the public evidence needed to call Grok 4.6 released. The model now has a tracked, unranked status page instead of a fabricated number. If you need an xAI coding model today, choose from the released models xAI currently documents and revisit the decision when 4.6 has a verifiable public profile.
