In general, OpenAI and Anthropic documentation lags their own products by months, and both companies have tons of obsolete and contradictory advice scattered around their sites.
RESEARCH
-
Using one model vs splitting tasks across models
By
–
Are you using the same model for everything or splitting tasks across models?
-
Publication bias: models learn only from successful science
By
–
This is publication bias inherited at full scale. Models learned from the sliver of science that worked and almost none of what didn't.
-
AI’s ERNIE model digitizes martial arts legacy, preserving ancient movements.
By
–
When ancient movements meet modern algorithms, AI carries on the legacy of martial arts.
— Baidu Research (@BaiduResearch) 4 juin 2026
Powered by Baidu ERNIE, the martial arts large model can capture movements, preserving and recording them in digital form.
This approach allows the traditional oral transmission and… https://t.co/6OSPtMm6huWhen ancient movements meet modern algorithms, AI carries on the legacy of martial arts. Powered by Baidu ERNIE, the martial arts large model can capture movements, preserving and recording them in digital form. This approach allows the traditional oral transmission and
-
AI hallucination leads candidate to wrong company with confidence
By
–
Somewhere, a candidate is walking into the wrong company with total confidence because their “research assistant” hallucinated the restaurant, the CEO, and possibly the menu. I laughed because this is exactly how AI confidence feels before verification enters the room. The
-
Hardmaru will discuss Sakana AI’s 1T parameter model on WBS
By
–
On TV Tokyo’s WBS (
@wbs_tvtokyo
) tonight I’ll be discussing Sakana AI’s upcoming 1T parameter model project, supported by METI’s GENIAC initiative. We are scaling up to build Japan’s first 1T parameter agent-native model, specifically optimized for long-horizon deep research and -
Tonight at 22:00 on WBS: GENIAC Selection Interview Broadcast
By
–
Broadcasting tonight at 22:00 on TV Tokyo WBS (@wbs_tvtokyo). We were interviewed about being selected for the METI AI development support project "GENIAC." Our company's CEO David Ha (@hardmaru) and Research Scientist Suginuma will discuss our strategy and the potential for
-
IndiaAI announces open-source foundational AI models for all
By
–
The foundational AI models being built under the IndiaAI Innovation Centre will be open-source — available to government organisations, startups, and researchers across the country. Follow IndiaAI for updates! #IndiaAI #IndiaAIMission #OpenSourceAI @AshwiniVaishnaw
-
Benchmark lets AI agents curate own training data without humans
By
–
What if an AI agent could curate its own training data—without human hand-holding? Researchers from Virginia Tech, UIUC, UW-Madison, and UC Berkeley present CURATION-BENCH, a benchmark that gives generalist coding agents command-line access to inspect, implement, test, and
-
Optimize the machine that produces model improvements, not just the next release
By
–
Most teams optimize for the next model release. Microsoft's MAI-Thinking-1 report argues for something different: optimize the machine that produces model improvements. The most valuable asset isn't a benchmark score. It's a training system that keeps getting better.