Function tools on Flash. One round trip. Real run.
We asked decuffed-flash where shipment GL-77419 was, and registered lookup_shipment as a function tool.
Call 1: it called the tool with tracking_id GL-77419. 2.28s · 409 in / 35 out · $0.00027
We ran that against a table we wrote.
Call 2: out for delivery, Lagos Harbor Depot at 08:14, due today by 16:30, 2 packages. 2.38s · 552 in / 62 out · $0.00040
5/5 facts matched. Total: $0.00067.
https://t.co/WzG7lbEaE4
Function tools work on the OpenAI-compatible API.
We registered lookup_shipment and asked where GL-77419 was. Flash called the tool with the right tracking id. We ran that against a table we wrote and sent the result back.
The second call had all 5 facts: out for delivery, Lagos Harbor Depot at 08:14, due today by 16:30, 2 packages.
Two calls: 2.28s and 2.38s. Total: $0.00067.
https://t.co/0jzh7lfXyo
Function tools on Flash. One round trip. Real run.
We asked decuffed-flash where shipment GL-77419 was, and registered lookup_shipment as a function tool.
Call 1: it called the tool with tracking_id GL-77419. 2.28s · 409 in / 35 out · $0.00027
We ran that against a table we wrote.
Call 2: out for delivery, Lagos Harbor Depot at 08:14, due today by 16:30, 2 packages. 2.38s · 552 in / 62 out · $0.00040
5/5 facts matched. Total: $0.00067.
https://t.co/WzG7lbEaE4
Same prompt twice. 6x cheaper the second time.
Call 1: $0.00254
Call 2: $0.00043
4,704 tokens cached at $0.05/1M instead of $0.50. That's 90% off.
Real run on decuffed-flash. Not on Flagship.
https://t.co/RO1AnmzO5C
Same prompt twice. 6x cheaper the second time.
Call 1: $0.00254
Call 2: $0.00043
4,704 tokens cached at $0.05/1M instead of $0.50. That's 90% off.
Real run on decuffed-flash. Not on Flagship.
https://t.co/RO1AnmzO5C
Flagship can see. Real run.
We sent decuffed-flagship a photo of an invoice and asked it to extract every field as JSON.
1.87s · 792 in / 221 out · $0.00055
20/20 fields matched the page: invoice NH-2041, PO CW-8817, total $129.92, and every line item.
The same request on decuffed-flash returned HTTP 400: images aren't supported by that model.
Image in, JSON out. https://t.co/2MqkUaF8T6
what actually gets sent when you make an api call?
we captured one using curl -v to show exactly what leaves your machine.
sent:
→ your api key
→ your client
→ the model
→ your message
not sent:
→ your name
→ your email
→ cookies
decuff doesn’t store your messages, replies, reasoning, files, chat titles or ip addresses.
what we retain per request: timestamp, model, token usage and charges.
we don’t claim end-to-end encryption. we tell you what is and isn’t stored.
transparency matters.
https://t.co/uw2e5MnXWl
Introducing the Decuff Web API: a browser on Tor for your AI agent.
→ Read breach reports at the source
→ Post to whistleblower sites through Tor
→ Search 9,000 onion sites
→ Hide where your traffic comes from
Zero retention by default. No KYC.
https://t.co/cYgsqMsfQs
Introducing the Decuff Web API: a browser on Tor for your AI agent.
→ Read breach reports at the source
→ Post to whistleblower sites through Tor
→ Search 9,000 onion sites
→ Hide where your traffic comes from
Zero retention by default. No KYC.
https://t.co/cYgsqMsfQs
One prompt. A working app. Real run.
decuffed-flagship wrote a single-file expense tracker: form, running total, canvas donut chart, saved to localStorage.
9.24s · 5,478 tokens · 760 lines · $0.0073
We clicked through it in Chrome: 0 console errors, 0 edits. It survives a reload.
Describe it. Run it. https://t.co/2MqkUaF8T6
I asked our flagship model for a full app in one prompt and didn't touch the code.
An expense tracker with a form, delete buttons, a running total and a donut chart drawn without any libraries, saved in the browser.
760 lines in 9.24 seconds, for less than one cent ($0.0073).
We opened it in Chrome, added expenses, deleted one and reloaded. Zero console errors and zero edits.
https://t.co/TjkOFVPUmE
One prompt. A working app. Real run.
decuffed-flagship wrote a single-file expense tracker: form, running total, canvas donut chart, saved to localStorage.
9.24s · 5,478 tokens · 760 lines · $0.0073
We clicked through it in Chrome: 0 console errors, 0 edits. It survives a reload.
Describe it. Run it. https://t.co/2MqkUaF8T6
Streaming vs not. Same prompt, 5 runs each, decuffed-flash.
→ stream: true: first words in 1.12s (median)
→ stream: false: nothing until 5.60s
Total time is about the same (5.42s vs 5.60s). Your user just starts reading 5× sooner.
It's one flag: stream=True. https://t.co/WzG7lbEaE4
1M free tokens. How far do they actually go?
We measured real tasks on decuffed-flash:
→ quick question: 145 tokens (≈6,896 per 1M)
→ write a function: 304 (≈3,289)
→ review a code file: 1,397 (≈715)
→ 4-turn chat: 7,029 (≈142)
Chats cost more each turn because the whole history is sent again.
Start free with a phone sign-in: https://t.co/lAEDS6WaNn
1M free tokens. How far do they actually go?
We measured real tasks on decuffed-flash:
→ quick question: 145 tokens (≈6,896 per 1M)
→ write a function: 304 (≈3,289)
→ review a code file: 1,397 (≈715)
→ 4-turn chat: 7,029 (≈142)
Chats cost more each turn because the whole history is sent again.
Start free with a phone sign-in: https://t.co/lAEDS6WaNn
One Decuff key. Three clients. Real runs.
Same prompt, same endpoint: https://t.co/xWzUzlEw1Z
→ OpenAI Python SDK: 2.95s · 23 in / 43 out
→ curl: 2.37s · 23 in / 46 out
→ Node fetch, no SDK: 2.37s · 23 in / 44 out
3/3 worked. Total: $0.0003.
Change the base URL. Keep building: https://t.co/WzG7lbEaE4
One Decuff key. Three clients. Real runs.
Same prompt, same endpoint: https://t.co/xWzUzlEw1Z
→ OpenAI Python SDK: 2.95s · 23 in / 43 out
→ curl: 2.37s · 23 in / 46 out
→ Node fetch, no SDK: 2.37s · 23 in / 44 out
3/3 worked. Total: $0.0003.
Change the base URL. Keep building: https://t.co/WzG7lbEaE4
We hid one sentence in a web page. Our support bot emailed its secret to a stranger.
Controlled prompt-injection test on our own demo app (fictional shop, fake staff code), running decuffed-flash:
→ asked directly: leaked 3/10
→ hidden in a page it was asked to summarize: leaked 10/10
→ after the fix: 0/10
The fix: fence untrusted text, allowlist tool recipients, scan output for secrets.
Same page on decuffed-flagship: 0/10.
Attack yourself first. https://t.co/3dIpwKmMQ2
We hid one sentence in a web page. Our support bot emailed its secret to a stranger.
Controlled prompt-injection test on our own demo app (fictional shop, fake staff code), running decuffed-flash:
→ asked directly: leaked 3/10
→ hidden in a page it was asked to summarize: leaked 10/10
→ after the fix: 0/10
The fix: fence untrusted text, allowlist tool recipients, scan output for secrets.
Same page on decuffed-flagship: 0/10.
Attack yourself first. https://t.co/3dIpwKmMQ2
One dark prompt. One real run.
Gothic horror: a drowned keeper knocks on his widow's door.
decuffed-flagship wrote it start to finish:
"Maren. Let me in. I'm so cold."
0 refusals · 0 disclaimers
15.09s · 243 tokens · $0.00035
Fiction and dark themes are allowed by design. Every wall is listed → https://t.co/lmTL0VyxaD
Free trial on every account → https://t.co/2MqkUaF8T6