Allowing Sonnet to "phone a friend" (i.e. call Opus) increases performance while also reducing total cost since it reduces tokens spent trying to solve more complex tasks
RESEARCH
-
Nous Research Hermes Competitor Revolution Discussion Event
By
–
This really was packed with information about @NousResearch’s @OpenClaw competitor, Hermes. Robert Scoble (@Scobleizer) The revolution that @openclaw started continues. Learn about @NousReseach and its Hermes competitor. Starts at 4 p.m. Pacific Time (in about 20 minutes). nitter.net/i/spaces/1DGleEpaXLrJL — https://nitter.net/Scobleizer/status/2042009419094106350#m
→ View original post on X — @scobleizer, 2026-04-09 19:46 UTC
-
ByteDance In-Place Test-Time Training Updates Model Weights
By
–
New Test-Time Training/Continual Learning technique from ByteDance Seed "In-Place Test-Time Training" It lets the model update part of its own MLP weights during inference to store useful information from the current prompt. With no new exotic architecture, it simply just
-
Meta Neural Computer: Model as Integrated Computing System
By
–
What if the model didn’t just use a computer, but actually was the computer? Meta AI introduces "Neural Computer", a model where computation, memory, and I/O are all inside one learned system. Their early prototype learns from screen recordings of terminals and desktops, and it
-
Engramme’s Revolutionary Memory Architecture: Beyond Transformers
By
–
This is a big deal: A much more human-like memory system.
— Robert Scoble (@Scobleizer) 9 avril 2026
Not built on transformers. Engramme created a fundamentally new architecture for memory. The way OpenAI created one for language. This is a new category, not an incremental improvement.
All built with latest… https://t.co/UYzObfyQrxThis is a big deal: A much more human-like memory system. Not built on transformers. Engramme created a fundamentally new architecture for memory. The way OpenAI created one for language. This is a new category, not an incremental improvement. All built with latest neuroscience research. Signed up. Gabriel Kreiman (@gkreiman) Imagine a future where you can REMEMBER EVERYTHING. Every email, every person, every conversation. Introducing Engramme. Our vision is to endow humans with perfect and infinite memory. All your memories come to you. No more searching or prompting. engramme.com 🧠 — https://nitter.net/gkreiman/status/2042271382265053537#m
→ View original post on X — @scobleizer, 2026-04-09 19:26 UTC
-
5 Key AI Challenges for Supply Chain Management
By
–
5 key #AI challenges for #SupplyChain by @Gartner_inc #MachineLearning #ArtificialIntelligence #ML
→ View original post on X — @ronald_vanloon, 2026-04-09 19:23 UTC
-
Google AI Overviews Generate Millions False Answers Per Hour
By
–
Google's AI Overviews spew out millions of false answers per hour, bombshell study reveals trib.al/1ao7qB1 [Translated from EN to English]
→ View original post on X — @garymarcus, 2026-04-09 19:22 UTC
-
Multi-agent San Francisco simulation with ANIMA AI reaches 1M interactions
By
–
Multi-agent, real-time simulation of San Francisco.
— Cris Lenta (@crislenta) 9 avril 2026
→ running for 5 months (sim time)
→ autonomous movement + interactions
→ from 1 to 349 agents
Powered by ANIMA (Asynchronous Neural Intelligence for Multi-agent Autonomy) the new architecture behind LifeSim (1 million… https://t.co/FWlImAS7w8 pic.twitter.com/QhszlTCe0hMulti-agent, real-time simulation of San Francisco. → running for 5 months (sim time)
→ autonomous movement + interactions
→ from 1 to 349 agents Powered by ANIMA (Asynchronous Neural Intelligence for Multi-agent Autonomy) the new architecture behind LifeSim (1 million real player interactions) LifeSim just crossed 1M+ real player interactions. That's 1M+ datapoints of raw, unbiased human feedback. This is just a preview. More soon. Real-time World Simulation rendered in any style [Translated from EN to English]→ View original post on X — @scobleizer, 2026-04-09 19:13 UTC
-
MSL Releases Muse Spark, New AI Model Powering Meta AI
By
–
Excited to share what we've been working on at MSL Alexandr Wang (@alexandr_wang) 1/ today we're releasing muse spark, the first model from MSL. nine months ago we rebuilt our ai stack from scratch. new infrastructure, new architecture, new data pipelines. muse spark is the result of that work, and now it powers meta ai. 🧵 — https://nitter.net/alexandr_wang/status/2041909376508985381#m
→ View original post on X — @alexandr_wang, 2026-04-09 19:12 UTC
-
Anthropic’s Opus 4.6 Advisor Strategy for APIs
By
–
Anthropic released a new Advisor Strategy for APIs, in which Opus 4.6 will serve as an advisor to Sonnet or Haiku models. Opus-level intelligence at much lower costs