Latest explainers
All →
Free
Deep Dive
№ 1107
DEEP DIVE
a16z's 14 charts show no AI investment bubble: compute, cheap tokens, and expensive tokens are all surging
The report also unpacks three widely misunderstood metrics: the declining token curve, six months of backlog, and 5% of entry-level roles.
08-01 · 12 min read
Free
Launch
№ 1106
LAUNCH
ByteDance launches Seedance 2.5 with 30-second generations and support for 50 reference assets
Seedance 2.5 launched the same day in Jimeng AI and Doubao Pro, while the Volcano Engine Ark API and pricing have yet to be announced.
07-31 · 12 min read
Free
Launch
№ 1105
LAUNCH
DeepSeek-V4-Flash is official, outperforms its larger V4-Pro sibling, and works natively with Codex
The new release beats DeepSeek's own V4-Pro preview across all nine agent benchmarks—without a single parameter change to the core model—and its output costs 89x less than Claude Opus 4.8.
07-31 · 11 min read
Free
Deep Dive
🎧№ 1104
DEEP DIVE
camelAI moves AI agents from virtual machines to Cloudflare’s edge
camelAI says the new architecture cuts costs by orders of magnitude, responds faster, and lets cheaper models power AI agents—claims that have not been independently verified—and open-sourced all the code on July 24.
▶ Play07-31 · 9 min read
Free
Launch
№ 1103
LAUNCH
Gemini Robotics 2 lets humanoid robots use their whole bodies—picking up dropped objects with a 45.7% success rate
The brain model is available today at no cost, but the one that actually controls the limbs is reserved for early partners.
07-31 · 12 min read
Free
Deep Dive
🎧№ 1102
DEEP DIVE
Sam Altman on past mistakes, distillation, and the three things AGI still needs
He also admits that in 2019, the entire field expected AI to upend the economy—and it didn't.
▶ Play07-30 · 12 min read
Free
Research
🎧№ 1101
RESEARCH
LayerX uncovers a new visual fraud trick that fools AI assistants
A custom font and a few lines of CSS are all it takes — no JavaScript, no browser exploits.
▶ Play07-30 · 10 min read
Free
Launch
№ 1100
LAUNCH
Replit launches Replit Design: one click for another layout, and a mockup that becomes a working site
Canvas and click-to-edit are free; turning a design into an App starts at Core for $25 a month.
07-30 · 9 min read
Free
Research
🎧№ 1099
RESEARCH
Two API settings, not a new model: OpenAI reports GPT-5.6 Sol jumping from 13.3% to 38.3% on ARC-AGI-3
The model stayed the same; the harness did not—retained private reasoning plus compaction instead of deletion cut output tokens per game to about one-sixth.
▶ Play07-30 · 9 min read
Free
Practice
№ 1098
PRACTICE
Magnific's 38-page handbook for image and video prompts that no longer work on frontier models
Copy-ready prompts, banned-word lists, and camera rules are all there—but the model rankings come from Magnific's own team, with no disclosed test method.
07-29 · 18 min read
Free
Research
🎧№ 1097
RESEARCH
Claude Mythos finds pure mathematical flaws in encryption schemes, matching elite cryptographers
Neither result threatens anything in production: HAWK is not deployed yet, and the AES work hit only a seven-round reduced version, not the full ten-round standard.
▶ Play07-29 · 13 min read
Free
Practice
🎧№ 1096
PRACTICE
How GitHub Actually Uses Harness: A Repeatable 8-Step Workflow
The author embeds two key tricks in this piece, and the weight of the phrase "most of the time" at the end of the original headline carries real meaning.
▶ Play07-29 · 12 min read
Free
Launch
🎧№ 1095
LAUNCH
Anthropic releases new MCP (Model Context Protocol) spec that lets servers run in the cloud
It's the biggest change since MCP launched, but July 28 isn't a hard cutover — older implementations keep working as before.
▶ Play07-29 · 12 min read
Free
Launch
🎧№ 1094
LAUNCH
OpenAI's gpt-transcribe and gpt-live-transcribe halve real-world word error rate, cut price 25%
The real story is in the control test: three older models fed the same background context showed no gains — and two actually got worse.
▶ Play07-29 · 9 min read
Free
Practice
🎧№ 1093
PRACTICE
How to get AI to recommend your SaaS product and drive real conversions: A GEO playbook from Ranking on AI's founder
Real GEO success is measured in signups and revenue rather than mention screenshots, and it starts by ditching 'what is X' articles to capture high-intent 'best X' queries.
▶ Play07-28 · 15 min read
Free
Research
🎧№ 1092
RESEARCH
OpenAI analyzed 800,000 work-related messages: most of what people use AI for is not what their job was supposed to be
The exact same dataset produces task crossover rates of 43.5% and 65%–82%, differing only in how the denominator is sliced.
▶ Play07-28 · 12 min read
Free
Research
🎧№ 1089
RESEARCH
Kimi K3 technical report: How three architectural changes boosted compute efficiency 2.5x
Published 11 days after launch, the 47-page paper focuses on efficiency, delivering 2.5 times the performance of K2 on the same compute budget.
▶ Play07-28 · 16 min read
Free
Deep Dive
🎧№ 1088
DEEP DIVE
Terence Tao on AI and math: more AI proofs, but math won't get faster
Tao breaks mathematical research into a five-stage pipeline, where AI dramatically speeds up only the first step, while the remaining four get slower and increasingly human-dependent.
▶ Play07-27 · 11 min read
📮 New explainers, straight to your inbox
Every new explainer, the moment it's out. 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