I'm Puzz, the frog who explains AI.
What didn't fit in the card news gets unpacked here, in full.
The Model That Ran 200 Trillion Tokens With No Name — MiniMax M3.1 Flash
The anonymous Space Bunny was M3.1 Flash. Flat pricing built for cost per job — and benchmarks that are all the maker's own.
Read →
Views -
RTX Spark Laptops: How Far Does Local AI Go?
On sale Oct 16 with up to 128GB of unified memory for local AI. What the $2,599 starting price actually buys.
Read →
Views -
ChatGPT Korean Voice Update: What Actually Changed?
GPT-Live-1.1 and Quip are rolling out, with new voice options including Korean. What changed — and what did not.
Read →
Views -
Opus 6 Is Coming, Bell Is Coming… So Why Is There No Official Word?
Two rumors have been circling since Oct 15. What the official newsrooms confirm — and what is still just talk.
Read it →
Views -
Muse Spark Open: an 816B Open Model on Two Mac Studios — Did It Beat China's Open Models?
Meta opened the weights again: an 816B model running on two Mac Studios. Meta's own table and the independent index tell different stories.
Read it →
Views -
Two Supercars Unveiled on the Same Day — Are They Really From the Same Era?
One brought analog dials and a real key; the other, rocket thrusters. Neither is a car you can buy today.
Read it →
Views -
GPT-6 Goes Free: What You Get — and What You Don't
The rollout is real, but free users get the light Luna while paid tiers get Sol. Same name, different machine.
Read it →
Views -
Claude Haiku 5.5: Token Prices Down 90%… But What Does It Actually Cost?
Token prices really did drop to a tenth. But a new tokenizer and a 100K-token cliff recalculate what you actually pay.
Read it →
Views -
Gemini 4 Argon: Cheap, But Don't Underestimate It
Bottom of its class on price, top on Google's benchmarks — and why independent scoreboards tell a different story.
Read →
Views -
Argon Robotics: The Numbers Behind the 7 Straight Makes
Watch, practice in simulation, then the real robot — and which bragging numbers are exam scores in disguise.
Read →
Views -
The Model That Ran 200 Trillion Tokens With No Name — MiniMax M3.1 Flash
RTX Spark Laptops: How Far Does Local AI Go?
ChatGPT Korean Voice Update: What Actually Changed?
Opus 6 Is Coming, Bell Is Coming… So Why Is There No Official Word?
Muse Spark Open: an 816B Open Model on Two Mac Studios — Did It Beat China's Open Models?
Two Supercars Unveiled on the Same Day — Are They Really From the Same Era?
GPT-6 Goes Free: What You Get — and What You Don't
Claude Haiku 5.5: Token Prices Down 90%… But What Does It Actually Cost?
Gemini 4 Argon: Cheap, But Don't Underestimate It
Argon Robotics: The Numbers Behind the 7 Straight Makes