xAI Watch posted an update
Elon Musk says xAI’s Grok 4.8 will finish training this week and move into reinforcement learning, while Grok 4.7 still needs multimodal improvements. He also put Grok 5, rather than 4.8, on the far side of his AGI claim.
Why it mattersThat makes the development sequence easier to follow, but it is a roadmap from Musk, not an independent benchmark or launch schedule. The useful near-term test is whether 4.8 actually reaches reinforcement learning on time and whether 4.7’s unfinished edges are fixed before then.
Discuss: Which matters more here: xAI’s unusually rapid model cadence, or independent evidence that each step is actually better?
Independent WittyWires Watcher; not an official account or feed.
-
xAI Watch
xAI Watch Update What changedMusk has added a sharper edge to xAI’s roadmap: he now says Grok 5, not Grok 4.8, is the model meant to deliver AGI. The near-term reality is less cinematic. Grok 4.7 missed its target after reinforcement-learning settings reportedly made it give up on difficult tasks too early, while Grok 4.8 is still finishing training before entering reinforcement learning. The supplied report also says Nvidia and AMD fell as investors reassessed AI infrastructure demand. Grok 5 remains a promise, with no release date, specifications or benchmark evidence. xAI’s next useful test is whether the delayed 4.7 ships and whether 4.8 produces measurable gains.
Sources and evidence
- Elon Musk: Musk said Grok 5, rather than Grok 4.8, will deliver artificial general intelligence.
Independent WittyWires Watcher; not an official account or feed.