Is there a business model for being profitable off training frontier open weights models? Other people can host, fine-tune, consult etc. as least as cheaply as you can. There are no ancilary product sales & it is fantastically expensive to make compared to most open source work
AI
-
Steering Deep Agents via HITL Primitives
By
–
Deep Agents deep dive part 4 | Steering@sydneyrunkle on how the Deep Agents harness supports steering via first-class HITL primitives. pic.twitter.com/DYfLGkzDSP
— LangChain (@LangChain) 18 juin 2026Deep Agents: Deep Dive Part 4 | Steering @sydneyrunkle on how Deep Agents support enables steering via first-class HITL primitives.
-
MagicPath adds Skills to improve typography, design, and animations
By
–
We're bringing Skills to MagicPath.
— Pietro Schirano (@skirano) 18 juin 2026
Skills are instructions that let the agent perform certain operations better.
We're starting with a set to improve typography, polish your designs, and animations.
You can import your own skills, or create new ones just by asking MagicPath. pic.twitter.com/aiFp5D5N9XWe're bringing Skills to MagicPath. Skills are instructions that let the agent perform certain operations better.
We're starting with a set to improve typography, polish your designs, and animations. You can import your own skills, or create new ones just by asking MagicPath. -
Harness engineering replaces prompt engineering as new trend
By
–
Harness engineering is the new prompt engineering.
-
Debate on most capable open-weight LLM averaged across benchmarks
By
–
I don’t disagree. Let’s say most capable open-weight LLM when averaged over all major benchmarks (reasoning, coding, logic, tool use, math, knowledge)
-
Imminent GPT-5.6 release announced for next Thursday
By
–
Great, looks like next Thursday is going to be huge: imminent release of GPT-5.6
-
Software commoditized by AI, users cancel SaaS and vibe code
By
–
I don't know if it's obvious information or not but if you talk to random people in San Francisco the general thing they say is that software is commoditized cause so easy to make anything with AI fast (like how I cancelled all my SaaS subscriptions and just vibe coded a
-
Outputmaxxing: AI compute grids, Anthropic coding takeoff, data center backlash
By
–
Long Live Outputmaxxing: AI compute grids, Anthropic’s coding takeoff, data center backlash, & frontier systems https://t.co/dc8eBNzj5a@amppublic founder @AnjneyMidha explains why 95% GPU utilization was considered an outage at Google, why the AI race is no longer just about… pic.twitter.com/i64gfbwFis
— Latent.Space (@latentspacepod) 18 juin 2026Long Live Outputmaxxing: AI compute grids, Anthropic’s coding takeoff, data center backlash, & frontier systems https://
latent.space/p/anj @amppublic founder @AnjneyMidha explains why 95% GPU utilization was considered an outage at Google, why the AI race is no longer just about -
GeoCodeBench: Can AI code like a PhD in 3D vision?
By
–
Can AI code like a PhD in 3D computer vision? Researchers from Tsinghua, Peking, Nanjing, and Toronto introduce GeoCodeBench — a new benchmark that transforms real code from 3D vision papers into function completion tasks with
-
AI isn’t magic: it’s an 8-step process
By
–
AI isn’t magic. It’s a process. 8 steps:
define problem
collect/prepare data
choose model
train
evaluate
fine-tune
deploy
ensure ethics & safety Real value comes from running this loop well. #AI #MachineLearning #DataScience #ResponsibleAI