NullVoider

Engineer | Building in the chaos.

[NullVoider @nullvoider07]

@elonmusk It's been a long time since a Grok model weights was made Open-source. When can we expect one?

[NullVoider @nullvoider07]

@miao_xiong_cs @SpaceXAI is hiring for the roles you are looking for. You can reach out @forwarddeploy or @xbxnxdxcxtx.

[NullVoider @nullvoider07]

@ClaudeDevs A great feature to collect data for Computer Use Agents training. 😏🤐

[SpaceXAI @SpaceXAI]

We've open-sourced Grok Build and have reset usage limits for all users.

Open sourcing Grok Build allows anyone to support making a reliable and robust harness. Check out our code, including the Git repo for the Grok Build CLI.

https://t.co/3SSvPu2Nrz

[NullVoider @nullvoider07]

@ayushjaiswal @sama 🤣🤣🤣

[NullVoider @nullvoider07]

@elonmusk Waiting for next iteration of grok after Grok 4.5. If it get better than Opus 4.8 then I will cancel my claude subscription and purchase annual supergrok subscription. 😃

[NullVoider @nullvoider07]

@ayushjaiswal OpenAI and Anthropic are being childish and fighting over 1 small piece of block from the Lego. 😂😂

[NullVoider @nullvoider07]

@ayushjaiswal Absolutely true. 🤣🤣

[NullVoider @nullvoider07]

@claudeai When will Fable 5 be available in claude Science?

[NVIDIA AI @NVIDIAAI]

We took a 30B model and split it in two to write tokens in parallel instead of one at a time.

Introducing Nemotron-Labs-TwoTower: a diffusion language model from NVIDIA Research adapted from Nemotron-3-Nano-30B-A3B. Here’s how it works: one half holds the context, the other writes the tokens, with both reusing the pretrained model instead of training a new one from scratch.

We found it kept 98.7% of the original model’s quality at 2.42× faster generation.

[Claude @claudeai]

Introducing Claude Science, a new app designed with every stage of research in mind.

Artifacts traced to their code, environments managed on demand, and 60+ optional scientific databases that you can connect.

Available now in beta.

[Elon Musk @elonmusk]

Grok 4.5, based on our 1.5T V9 foundation model, with Cursor data added in supplemental training, is now in private beta at SpaceX & Tesla. Early evals show performance close to, perhaps exceeding Opus.

RL is continuing to significantly improve the model, and the Grok Build harness gets better every day.

Nice work by all those involved!

Completely trained from scratch new models will be released by @SpaceX every month this year.

[NullVoider @nullvoider07]

@elonmusk No need to rush with 1 model per month maybe 1 every 2 month should be right approach. Rushing will leave gaps in model capabilities. Take enough time and care to test everything just like how SpaceX test their rockets because it's mission critical. 😊😊

[NullVoider @nullvoider07]

@JasonBud @XFreeze How about chips designing tools? It can help designers and developers during the chip designing process.

[NullVoider @nullvoider07]

@ojhaayush @grok Grok build has more useful features and is better than claude code but the grok itself is still not as good at coding as claude. That's the main reason most people haven't switched to grok build yet.

[NullVoider @nullvoider07]

@IamEmily2050 When you have too much resources you tend to delay the work, but when you have way too low resources you end up finding ways to create something great even with the small amount of resources. 😊

[NullVoider @nullvoider07]

@aarnogau @justoutquan If development continued giving on enterprise a lot then it's pointless because everything will keep revolving in in circles.

[NullVoider @nullvoider07]

@ayushjaiswal Human not in the loop is a serious issue.

[NullVoider @nullvoider07]

@trikcode This is why when you work with help of these models you have each of code. Human in the loop is a must if you want to understand what's going on. 😄

[NullVoider @nullvoider07]

@gauravisnotme How are these a topic to flex upon? Whoever does it is really gonna get to an hospital before long. 🤣🤣