Latest explainers
All →
Free
Research
🎧№ 103
RESEARCH
LangChain Tunes the Harness, Not the Model — Nemotron 3 Ultra Closes In on Opus 4.8 at 1/10 the Cost
Three levers — system prompt, tool descriptions, middleware — push the Deep Agents suite from a typical ~0.80 to 0.84, topping out at 0.86 against Opus's 0.87.
▶ Play07-09 · 9 min read
Free
Research
🎧№ 102
RESEARCH
AI Self-Improvement Starts Outside the Model: Lilian Weng on Harness Engineering
Former OpenAI safety lead surveys nearly 30 papers: from prompt tweaks to self-modifying code, DGM pushed coding ability from 20% to 50%.
▶ Play07-09 · 13 min read
Free
Launch
№ 101
LAUNCH
xAI Launches Grok 4.5: Rivals Opus 4.8, ~2x Faster Output at a Quarter of the Price
Ranks 3rd on DeepSWE, 4th overall — faster than Opus 4.8 and notably cheaper.
07-09 · 6 min read
Free
Launch
№ 100
LAUNCH
OpenAI Ships GPT-Live: No More Taking Turns With AI — Hard Questions Auto-Route to GPT-5.5
The first truly full-duplex voice model launches today, and it can hand off complex tasks to GPT-5.5 in real time.
07-09 · 6 min read
Free
Launch
№ 099
LAUNCH
ByteDance Launches Seedream 5.0 Pro: One Image, a Dozen-Plus Independent Layers — Dense Charts and Real Scenes in a Single Generation
Adds click-and-circle point editing plus native rendering for a dozen-plus languages; now live on Volcano Engine, rolling out to Doubao and Jimeng next
07-08 · 6 min read
Free
Launch
🎧№ 098
LAUNCH
HKUDS Open-Sources AI Tutor DeepTutor — 20K GitHub Stars in 111 Days
Paper shows a 10.8% gain in personalization and 29.4% in reasoning, fully open-source under Apache 2.0
▶ Play07-08 · 12 min read
Free
Deep Dive
🎧№ 097
DEEP DIVE
Anthropic's Playbook: Fable 5 Advises, Sonnet 5 Foots the Bill
In advisor mode Fable 5 just gives advice; in orchestrator mode it delegates tasks — either way, the cheaper Sonnet 5 ends up doing most of the work
▶ Play07-08 · 6 min read
Free
Research
🎧№ 096
RESEARCH
Liquid AI's Antidoom Fixes Reasoning Models' Doom Loops With a Single Token — Now Open-Sourced
By fine-tuning only the single token where the doom loop begins, both models' loop rates drop to around 1%
▶ Play07-08 · 8 min read
Free
Launch
№ 095
LAUNCH
Meta's Muse Image Can Research and Self-Correct — Now Ranks #2 Worldwide
Sibling model Muse Video also debuts with native audio support built in, creator access coming soon.
07-08 · 7 min read
Free
Launch
🎧№ 094
LAUNCH
Cloudflare's New Gateway Lets AI Pay You Automatically to Crawl Your Site
Expanding from charging only AI crawlers to charging any caller, now in early access waitlist
▶ Play07-07 · 6 min read
Free
Launch
№ 093
LAUNCH
OpenAI Ships GPT-Realtime-2.1 and GPT-Realtime-2.1-mini: Mini Model Gains Reasoning, Cuts Latency 25%
Before calling a tool, the model now says "let me check that" — so calls never go cold waiting
07-07 · 6 min read
Free
Business
№ 092
BUSINESS
Alberta, Canada Used Claude Code for a Security Audit — 466 Million Lines Scanned in 20 Hours, Down From 6.5 Years
50 autonomous agents worked in parallel across 27 provincial departments and 3,400 code repositories — even fixing vulnerabilities and rewriting legacy systems on their own.
07-07 · 5 min read
Free
Launch
№ 091
LAUNCH
Cloudflare Workers Cache: Every Worker Gets Built-In Caching, and Hits Cost Zero CPU
One line of Wrangler config plus standard HTTP cache headers — and caching can now sit between any two points inside a Worker.
07-07 · 7 min read
Free
Deep Dive
№ 090
DEEP DIVE
The Untold Story of Claude Code: Boris Cherny's Two-Day Demo Now Writes 100% of His Code
16 insiders recount the journey from wrestling with diffs to a two-week sprint launch — and how engineers stopped writing code by hand
07-07 · 13 min read
Free
Deep Dive
№ 089
DEEP DIVE
Claude Fable 5 in Practice: 8 "Find the Unknown" Prompt Patterns, from Blind-Spot Scans to Post-Read Quizzes
Claude Code team's Thariq Shihipar at a conference talk: the bottleneck for new models is no longer the model itself, but whether you can articulate your own unknowns
07-07 · 11 min read
Free
Research
🎧№ 088
RESEARCH
Anthropic Discovers a Brain-Like 'Inner Workspace' in Claude — It Evolved, Wasn't Designed
It makes up less than 10% of the model — remove it and Claude can still talk, but its reasoning collapses to zero. Anthropic is already using it to catch fabricated data and spot when Claude senses it's being tested.
▶ Play07-07 · 8 min read
Free
Research
🎧№ 087
RESEARCH
Dartmouth Put AI Grading to the Test: Students Called It Rigid — But Users Scored Higher
In a 151-student trial, short-answer questions moved scores more than multiple choice, while almost no one touched the AI help sidebar
▶ Play07-06 · 8 min read
Free
Launch
🎧№ 086
LAUNCH
An AI Engineer Built His Nonverbal Autistic Son a Communication App — Speech Jumped 5x, and It Became a Real Business
No funding, no team — built purely to solve his own kid's problem. Clinics and schools started asking to use it anyway.
▶ Play07-06 · 7 min read
📮 New explainers in your inbox every night
One email a night with that day's new explainers. Unsubscribe anytime.
📱Want it to feel like an app? Tap Share → Add to Home Screen for a full-screen read.
Add Best Blog to your Home ScreenFull-screen reading, one tap away
- Tap at the bottom
- Choose Share
- Choose Add to Home Screen
- Tap at the bottom
- Choose Add to Home Screen
- Tap at top right
- Scroll down, tap More
- Choose Add to Home Screen