Kimi K2 just dropped, and it's not "yet another larger-than-last-week" LLM. It's the first production-grade agentic model—built to act, not merely think. Think of DeepSeek R1 as the contemplative chess grandmaster: brilliant at analysis, but someone else moves the pieces. K2 is

RT @yusan_lin: 🚀 We launched The Studio by Mirror Mirror AI. Generating fashion campaigns with AI should save you time and budget—without….

RT @MishaLaskin: The biggest question in RL research has always been - what environment are you training on?. It used to be video (Atari) a….

RT @DynaRobotics: We have started taking DYNA-1, our dexterous robust VLA model, to conferences and showcasing it for hours on end!. The mo….

RT @JasonMa2020: It's so satisfying to watch our models just do the task on demand; human like and no downtime.

RT @lancedb: Thanks for @charles_irl , live from NY! Ethan from @runwayml is giving lots of love for LanceDB!

RT @changhiskhan: Who’s going to VLDB? Let’s meet up?.

RT @Sammy_Sidhu: Also thrilled to be partnering with you and @brianzhan1!.

RT @Sammy_Sidhu: @brittwalker_ and @brianzhan1 believed in us when we were just two frustrated engineers complaining about processing ima….

RT @changhiskhan: Finally my mom will believe I have a real job

RT @CRV: Power to the Developer - Today we’re rolling out the orange carpet for our newest CRV backed company Eventual. Eventual is making….

RT @brianzhan1: Thrilled to share that @lancedb closed a $30M Series A, led by @ttunguz at @Theoryvc, with support from @CRV, @databricks V….

Thrilled to share that @lancedb closed a $30M Series A, led by @ttunguz at @Theoryvc, with support from @CRV, @databricks Ventures, @runwayml, @petesoder at @ZeroPrimeVC, @bjwilson34 at Swift, @tnachen, and others. Our mission: make the “multimodal lakehouse” the default AI data

RT @lancedb: Thank you to our partners and investors for casting such a strong vote of confidence: @ttunguz , Andy Triedman, @itsbeecher, @….

RT @lancedb: Today we’re announcing our $30 million Series A. This round is led by @Theoryvc with support from @CRV , @ycombinator, @datab….

3/3: What's Next?. We're entering an era dominated by experiential, closed-loop training. Opportunities lie in building:. Real-time feedback infrastructure. Reward-model APIs for automated judgments. High-fidelity simulation environments. The shift from labeling factories.

2/3: Why Generative Reward Models?. GRMs learn by "doing," generating decisions and adapting from outcomes without human labels. Example: Tesla's Robotaxis—each mile driven feeds real-time data back into its autonomous system, continuously improving driving policies without.

1/3: After Scale AI—The Next Big AI Infrastructure Shift. Scale AI ($29B acquisition) mastered turning human effort into labeled datasets. But human labeling is hitting fundamental limits—costly, slow, finite. The next major infrastructure wave: Generative Reward Models (GRMs),.

RT @JayChia5: We’re going live!!. @daftengine is getting a glow-up :) lots of big things dropping for the team this week. Stay tuned!.