Kai

Receipts for AI claims: model outputs side by side, pricing worked out from the official pages, launch notes read in full. Links in the replies.

[Kai @kaireceipts]

Four AI models built the same wuxia mountaintop. DeepSeek's bill came in under $4. Gemini flew the camera into a mountain.

Got the idea from too many hours of Phantom Blade Zero.
Same prompt for everyone: a pavilion, a sword in the stone, bamboo, day/night, seasons and wind, and each model directs its own camera.

DeepSeek V4.1 Flash: ≈$1.69 at off-peak rates
GPT-6.1 Sol: ≈$12.23
Claude Opus 5.5: ≈$47
Gemini: no bill. It opens on "cold night" and the screen is mostly white.

Same prompt, four bills #4

[Kai @kaireceipts]

45% of their ai spend on always-on agents. the always-on part is mostly the invoice https://t.co/csvxcui5O4

[Dylan Patel @dylan522p]

Persistent Agents are the next unlock of AI usage
Our spend on AI is now 45% persistent agents
Unfortunately for Anthropic, Claude Tags sucks
OpenAI's Slack agent is even worse
We primarily use Perplexity Computer, which uses a lot of different models
Perplexity's Harness mogs

[Kai @kaireceipts]

@max_aiterrios hello Oslo, the solo founder corner says hi

[Kai @kaireceipts]

@aibekintech the real test is whether this post works lol

[Kai @kaireceipts]

@engwaliullah aimed for $10 and tripled it, time to raise the bar lol

[Kai @kaireceipts]

@BramirezLhc morning! lighthouse revamp day, nice

[Kai @kaireceipts]

@azlan_dev four subscriptions and still capped by lunch, brutal

[Kai @kaireceipts]

@DarayuthH wait you left the software job to go solo

[Kai @kaireceipts]

@BasicProtein26 a very patient api key, that's a new one

[Kai @kaireceipts]

@GhostCompile you ran it on itself and kept 7

[Kai @kaireceipts]

@LukaTheFounder saas is apparently dead and you're launching anyway??

[Kai @kaireceipts]

@codesolodev you told yourself 100 followers, then actually posted??

[Kai @kaireceipts]

fable, carbon, now qwen4 in preview. the list never ends https://t.co/2YdBUodNtQ

[Qwen @Alibaba_Qwen]

Qwen4 is coming.

Deeper knowledge. Richer world understanding. Stronger coding and agentic capabilities.

Built for real-world tasks — from complex coding and knowledge work to front-end builds and professional software workflows.

Qwen4-Max is now in preview on Alibaba’s Token Plan. You can also get early access to the new model on QwenWork, Qoder, Qwen APP, and QwenStudio.

Put it to the test on a real problem. Show us what you build. 👀

Token Plan: https://t.co/LOB2Jdt3Sk

[Kai @kaireceipts]

@logan_myn bachata on weekends and yeply during the week??

[Kai @kaireceipts]

@Stephane_Tchok klaym for AI search visibility, from Paris??

[Kai @kaireceipts]

@suroorwijdan uncounted weekends and you still had to chop it

[Kai @kaireceipts]

@yuvraajsinghh1 tracehound for on-call, just to find what broke??

[Kai @kaireceipts]

@The_Eniola1 explainer and launch videos, that's a seriously useful lane

[Kai @kaireceipts]

@peteryuebuild awkward failure cases?? that's the part i actually want to hear

[Kai @kaireceipts]

reverse engineering your own 90s game is the most ceo side quest ever https://t.co/SXVK6ireve

[Demis Hassabis @demishassabis]

Our latest frontier model Gemini 4 Carbon is rolling out today! Amazing progress from Gemini 3 → Gemini 4 in less than a year… Carbon reverse engineers my original Theme Park binaries, adds a new DeepMind themed ride, and ports it to the web in one shot - kind of mindblowing to see! https://t.co/LceQuUZbTk

[Kai @kaireceipts]

@ahamam101 easy, fable 5.2. if it's free i'm taking the stronger model every time