Musk maps Grok 4.8, 4.9 and 5 while 4.7 is still late — and while he backs pacing
4.8 is a 2.5T run on a new C++ stack, finishing training this week then RL. 4.9 is “probably Astra/Fable class.” Asked about AGI, he pointed to Grok 5. xAI’s docs are silent.
Feature1 minIn-depth
James Okonkwo1 min read
- xAI
- Grok
- Grok 5
- Musk
- Models

Photo: TeslaNorthSource page
Elon Musk laid out xAI’s next few Grok rungs on X on September 14, TeslaNorth reported. Asked what else was coming in September besides Grok 4.7, he wrote: “Grok 4.8, which is a 2.5T model trained with our new C++ software stack, will finish training this week and start RL.” Asked how close 4.8 would be to AGI, he answered, “That will be Grok 5.” A longer follow-up put 4.7 “roughly on par with Opus 5.0, not 5.1,” called 4.8 “a noticeable improvement,” said 4.9 is “probably Astra/Fable class,” and left Grok 5 as “maybe better than anything. We shall see.” None of the posts named a Grok 5 date or a public scoreboard. xAI’s documentation, TeslaNorth noted, still has not restated the ladder.
Grok 4.7 should be roughly on par with Opus 5.0, not 5.1. Better in some ways, worse in others. We need to fix multimodal performance. Grok 4.8 will be a noticeable improvement. Grok 4.9 is probably Astra/Fable class. Grok 5 maybe better than anything. We shall see.
— Elon Musk, posts on X, Sep 14, 2026, as quoted by TeslaNorth
Grok 4.7 itself is still unreleased after a September 12 target. Musk had already said it needed more reinforcement learning — including a length penalty that made the model give up on long work. The new posts sit on top of that slip: 4.7 is still cooking; 4.8 is already described as finishing pretrain this week.
The same account spent the weekend agreeing with Dario Amodei. Musk quoted Anthropic’s evaluator announcement and wrote “Dario is right,” then added that “peer review of AI by competitors is the right way to start this off.” TeslaNorth set the ladder next to that endorsement without calling it a contradiction. The public sequence is still: finish 4.8, treat 4.9 as the competitive rung, reserve Grok 5 for the AGI-shaped claim. Oversight is a sentence. The training calendar is a list.


