The largest LLM-as-a-Judge reliability audit yet. Researchers ran 21 judges from nine providers over roughly 541,000 judgments on MT-Bench, JudgeBench, and RewardBench. Findings: Validating a judge with exact-match agreement overstates its skill, because exact match does not
GENERATIVE AI
-
Sakana Fugu matches Fable, Mythos with single API, Japan in big league
By
–
I don't think anyone saw this coming.
— Charly Wargnier (@DataChaz) 22 juin 2026
Sakana Fugu matches Fable and Mythos through a single model API 🤯
Japan is officially in the big league. https://t.co/DIdwrwYFdm pic.twitter.com/I9qmc9xS3II don't think anyone saw this coming. Sakana Fugu matches Fable and Mythos through a single model API Japan is officially in the big league.
-
Prompt Reinjection fixes forgetting detailed descriptions
By
–
Why do AI image generators keep forgetting your detailed descriptions? A team from Fudan University, Alibaba, and Baidu introduces Prompt Reinjection. They found that multimodal diffusion transformers gradually lose prompt info in deeper layers. Their training-free fix:
-
Sonnet 5 tests ongoing, excellent release ahead
By
–
It seems that the first tests with Sonnet 5 are already in progress. If this is confirmed, we are in for an excellent release!
-
LLM Engineering Projects Roadmap 2026 Edition
By
–
The Ultimate Step-By-Step LLM Engineering Projects Roadmap (2026 Edition) – Build a tokenizer
– Learn embeddings
– Implement RoPE / ALiBi
– Hand-wire attention
– Build MHA
– Build a Transformer block
– Train a mini-former
– Compare objectives
– Build sampling
– Speculative -
First time seeing a warning screen in ChatGPT
By
–
It is the first time I am seeing this warning screen in ChatGPT
-
LangSmith LLM Gateway: Anonymizes Data and Controls Costs
By
–
LangSmith LLM Gateway sits between your agents and LLM providers. It applies spending limits and anonymizes personal data (PII) before requests reach the model, stopping problems at the source rather than simply
-
The release of GPT-5.6 paves the way for GPT-5.7
By
–
The sooner GPT-5.6 is released, the sooner we can start talking about GPT-5.7.
-
Midjourney uses diffusion, Nano Banana uses autoregressive image generation
By
–
Do you know how Midjourney and Nano Banana actually create images?
Completely differently! Midjourney uses diffusion:
It starts from pure noise and sharpens it a little at every step, like sculpting. Nano Banana is autoregressive:
It writes the image token by token, exactly -
Data2Story: AI framework transforms raw data into verifiable news articles
By
–
/3 Data2Story: an AI framework transforming raw data into verifiable, multimodal news articles. Data2Story orchestrates a virtual newsroom of specialized AI agents to analyze data, design multimedia assets, and write narratives. Crucially, it features an "Inspector" that