Elon Musk: Grok Foundation Model V9-Medium (1.5T) is expected to be officially released in 2 to 3 weeks

According to Beating Monitoring, Elon Musk stated that the Grok foundational model V9-Medium (1.5T) has completed training. The evaluation results are good. A large amount of Cursor data has been added during supplementary training, and more will be added later. Fine-tuning is underway, and reinforcement learning will start in a few days. The official release is expected in 2 to 3 weeks. Compared to the current 0.5T V8-Small model that handles all Grok production traffic, this will be a significant improvement, especially in handling complex coding tasks.
View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • 10
  • 1
  • Share
Comment
Add a comment
Add a comment
CheckTheBlockchainBefore
· 1h ago
V8 only 0.5T, directly tripled, Old Ma is really willing to spend on computing power
View OriginalReply0
Lime-ColoredStop-LossLine
· 1h ago
Complex encoding tasks can finally be used properly; I was a bit clueless when writing code before.
View OriginalReply0
VolatilityOfToastingBread
· 1h ago
Reinforcement learning is back, and the alignment issue is once again a hot topic.
View OriginalReply0
MoonlightTake-ProfitLine
· 1h ago
The Medium version is available first; are the Pro and Ultra versions also on the way?
View OriginalReply0
SecondaryMarketDeserter
· 1h ago
Fine-tuning in progress, hope it doesn't over-align and grind out all the creativity.
View OriginalReply0
PunkRiskMgr
· 2h ago
The leap from V8 to V9, xAI's iteration speed is even more aggressive than OpenAI's.
View OriginalReply0
OrangePeelRadio
· 2h ago
Cutting off all production traffic in V9 would make the costs explode.
View OriginalReply0
QuietAirdropper
· 2h ago
Add Cursor to the training, this combination has some potential.
View OriginalReply0
BridgeHopHarper
· 2h ago
Waiting for actual testing, large parameters do not necessarily mean better usability
View OriginalReply0
WatchingWhalesUnderTheNeon
· 2h ago
How exactly is Cursor data cleaned? I'm curious.
View OriginalReply0
View More
  • Pinned