AI Explained
LLM benchmarks, research papers, and model releases with a focus on technical scrutiny and system card analysis.
Nutrition Label
AI Explained provides high-signal analysis of the latest AI developments, prioritizing academic papers and system cards over press releases. The content excels at identifying technical limitations and 'hidden flaws' in new models through independent benchmarking and close reading of documentation. Viewers receive a skeptical, evidence-based perspective, though the depth of hands-on demonstrations varies compared to the theoretical analysis.
Strengths
- +
- +
- +
Notes
- !Check descriptions for disclosures regarding the creator's own benchmarking platform and software products.
- !Analysis often relies on the creator's proprietary 'Council' benchmarks rather than standard public metrics.
Rating Breakdown
Breakdown across the key dimensions we rate. Methodology →
Recent Videos
What the New ChatGPT 5.4 Means for the World
Deadline Day for Autonomous AI Weapons & Mass Surveillance
Gemini 3.1 Pro and the Downfall of Benchmarks: Welcome to the Vibe Era of AI
The Two Best AI Models/Enemies Just Got Released Simultaneously
Claude AI Co-founder Publishes 4 Big Claims about Near Future: Breakdown
Anthropic: Our AI just created a tool that can ‘automate all white collar work’, Me:
What the Freakiness of 2025 in AI Tells Us About 2026
Gemini Exponential, Demis Hassabis' ‘Proto-AGI’ coming, but …
GPT 5.2: OpenAI Strikes Back
You Are Being Told Contradictory Things About AI
Nano Banana Pro: But Did You Catch These 10 Details?
Gemini 3 Pro: Breakdown
Is GPT-5.1 Really an Upgrade? But Models Can Auto-Hack Govts, so … there’s that
Bubble or No Bubble, AI Keeps Progressing (ft. Relentless Learning + Introspection)
Did you miss these 2 AI stories? A *Real* LLM-crafted Breakthrough + Continual Learning Blocked?
Why this rating
Evidence receipts showing why each dimension is rated the way it is.
“Nano Pro also available on my solo-made app (Pro Tier): https://lmcouncil.ai”[Description] →
Creator explicitly discloses his own commercial product (LMCouncil) and the video sponsor (AssemblyAI) in the description and video flow.
“I'm going to upload the entire Three.js library... that is 100,000 tokens.”[05:12] →
The creator demonstrates direct, first-hand engagement by uploading a massive custom dataset to test the context window live, rather than relying on press materials.
“I tested the needle in a haystack... and it found the phrase perfectly.”[02:45] →
Conducts original benchmarking (Needle in a Haystack) to verify the 1-million-token claim, providing concrete evidence of the model's recall capabilities.
“Gemini 1.5 Pro is out...”[00:01] →
The video content and audio explicitly identify the subject as 'Gemini 1.5 Pro', which contradicts the provided title 'Gemini 3 Pro', constituting a significant accuracy failure in the metadata.