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
LLMS
-
Hardmaru will discuss Sakana AI’s 1T parameter model on WBS
By
–
-
Anthropic co-founder: self-aware superhuman coder with nation-state hacking
By
–
6. Co-founder @jackclarkSF tells new recruits in a monthly welcome talk titled “Anthropic is not a normal company”: “we’re selling a superhuman coder that is increasingly self-aware and has nation-state hacking capabilities. Get hobbies that aren’t computers.”
-
Official MiniMax AI M2.7 documentation and API links
By
–
… and for your reference, here is the official @MiniMax_AI documentation: – MiniMax M2.7 official page: https://
minimax.io/models/text/m27
– MiniMax API docs: https://
platform.minimax.io/docs/api-refer
ence/api-overview
… -
SambaCloud’s MASSIVE 435 t/s output speed for OpenClaw multi-agent frameworks
By
–
When you run multi-agent frameworks like @OpenClaw
, you know the output speed is critical. As the chart below shows, SambaCloud serves MiniMax M2.7 at a MASSIVE 435 output tokens per second, more than 3x faster than the nearest competitor (Fireworks at 127 t/s). Combine that -
Gary Marcus’s 3 Predictions: Neurosymbolic AI’s Fortune, LLMs’ Limited Profitability
By
–
Three Predictions: 1. Some form of AI, probably neurosymbolic in nature, will come that is far more economical and data- and energy-efficient than LLMs, and it will make an absolute fortune. 2. LLMs, on the other hand, will never be all that profitable (aside from the chip
-
Trust not intelligence is the real problem with AI investing
By
–
Everyone wants “smarter AI” in investing.
— Linus ✦ Ekenstam (@LinusEkenstam) 4 juin 2026
Cute.
The problem was never intelligence.
It was trust.
Because “mostly right” sounds acceptable until your $50M acquisition is built on one hallucinated number.
That’s the gotcha.
AI can be brilliant and still be unusable when the… https://t.co/mdeciHCkW0Everyone wants “smarter AI” in investing. Cute. The problem was never intelligence. It was trust. Because “mostly right” sounds acceptable until your $50M acquisition is built on one hallucinated number. That’s the gotcha. AI can be brilliant and still be unusable when the
-
AI Answers Beat Law Professors in Blind Stanford Study
By
–
A blind Stanford-led study of nearly 3,000 anonymized matchups found law professors across 16 schools preferred AI-generated answers to student contract-law questions over those written by fellow professors 75% of the time, and judged the AI responses far less likely to be
-
Ask Codex to find the single most valuable time-saving product for you
By
–
This might be the craziest Codex workflow I've found.
— CHOI (@arrakis_ai) 4 juin 2026
Ask ChatGPT or Codex:
"Based on everything you've learned from our conversations so far, identify the single most valuable service, product, or AI-powered system that could reduce my working time by more than 50%."
Then… pic.twitter.com/8XmbjmODPxThis might be the craziest Codex workflow I've found. Ask ChatGPT or Codex: "Based on everything you've learned from our conversations so far, identify the single most valuable service, product, or AI-powered system that could reduce my working time by more than 50%." Then
-
Researchers propose CODA to keep data on chip longer for AI training
By
–
Can AI training be fixed by keeping data on the chip longer? Researchers from MIT, Princeton, Together AI, and Meta introduce CODA — a new way to rewrite Transformer building blocks as GEMM-plus-epilogue programs. Instead of moving large intermediate tensors back and forth to
-
Hermes Agent splits tasks between local agents and codex/claude code
By
–
I talked about a similar setup for coding Agents. Hermes Agent can split the task between local agents and codex/claude code. Step-by-step guide: