OpenGradient
102 Publications
OpenGradient
Compte Square Vérifié+The Network for Open Intelligence. Host models, run secure inference, and deploy agents verifiably onchain. Backed by a16z crypto, CB Ventures, and more.
0 Suivis
46.5K+ Abonnés
5.4K+ J’aime
Publications
Here's how to generate a 30-second clip using Seedance 2.5
→ On Image → Video, upload your reference images
→ Select an aspect ratio and a 30-second duration
→ Add up to 10k character prompt
→ Click the generate button
That's it!
Try it here → https://chat.opengradient.ai/video
→ On Image → Video, upload your reference images
→ Select an aspect ratio and a 30-second duration
→ Add up to 10k character prompt
→ Click the generate button
That's it!
Try it here → https://chat.opengradient.ai/video
Got offered unlimited Seedance 2.5?
Unlimited just means unlimited until it starts costing them money.
Ask anyone who paid for a year upfront.
Part 8: The AI Talk
Unlimited just means unlimited until it starts costing them money.
Ask anyone who paid for a year upfront.
Part 8: The AI Talk
Real life GTA before GTA 6 😅
Seedance 2.5 cooked this one.
Try it yourself https://chat.opengradient.ai
Seedance 2.5 cooked this one.
Try it yourself https://chat.opengradient.ai
Seedance 2.5 is live in the OpenGradient Video Studio. 30 seconds in one generation, one prompt, sound included. We wrote up what it's good at, what it costs, and what still isn't there. http://x.com/i/article/2086757629309956097
Private AI on Base 🟦
Seedance 2.5 is now LIVE on OpenGradient!
This entire film was generated from a single prompt. One take, no edits.
→ Up to 30-second clips from one prompt
→ 10,000 character prompts for full scene-by-scene direction
→ Up to 30 reference uploads for character and environment
This entire film was generated from a single prompt. One take, no edits.
→ Up to 30-second clips from one prompt
→ 10,000 character prompts for full scene-by-scene direction
→ Up to 30 reference uploads for character and environment
Seedance 2.5 is coming to OpenGradient!
Generate up to 30-second clips with 5k character prompts, up from 2k
Up to 50 multimodal references for character and environment consistency
This entire clip was made from 1 prompt, with no edits!
Register now→ http://chat.opengradient.ai
Generate up to 30-second clips with 5k character prompts, up from 2k
Up to 50 multimodal references for character and environment consistency
This entire clip was made from 1 prompt, with no edits!
Register now→ http://chat.opengradient.ai
Seedream 5.0 is live in OpenGradient Image Studio.
ByteDance's newest and strongest image model, now private and uncensored by default.
Sharp 2K detail, clean design, and text that actually renders right, even engraved on a tennis ball.
Try it: http://chat.opengradient.ai/image
ByteDance's newest and strongest image model, now private and uncensored by default.
Sharp 2K detail, clean design, and text that actually renders right, even engraved on a tennis ball.
Try it: http://chat.opengradient.ai/image
2 billion processed tokens, and it is growing fast!
That is billions of real questions and answers, and everything is private.
438,000 prompts have run through OpenGradient, and not one ever leaked, not even to us.
Every request is sealed inside a hardware enclave and
That is billions of real questions and answers, and everything is private.
438,000 prompts have run through OpenGradient, and not one ever leaked, not even to us.
Every request is sealed inside a hardware enclave and
The Moonpay x Claude/ChatGPT integration is a great innovation.
But if you are conservative and don't want anyone to see your trades, don't connect your AI.
Keep it private.
But if you are conservative and don't want anyone to see your trades, don't connect your AI.
Keep it private.
Your Claude or ChatGPT can use your wallet now?
So the people behind them see everything you buy, including the trades you would rather forget.
Never give a model that much of you when it trains on your data or sells it.
Go Private http://chat.opengradient.ai
Part 7: The AI Talk
So the people behind them see everything you buy, including the trades you would rather forget.
Never give a model that much of you when it trains on your data or sells it.
Go Private http://chat.opengradient.ai
Part 7: The AI Talk
Private AI was supposed to be impossible to prove. OpenGradient Chat made it verifiable.
Same frontier models you already use: GPT, Claude, ByteDance, Nous Research, Gemini, Grok, etc...
Ask anything, privately → http://chat.opengradient.ai
7/
Same frontier models you already use: GPT, Claude, ByteDance, Nous Research, Gemini, Grok, etc...
Ask anything, privately → http://chat.opengradient.ai
7/
Partiellement vrai
And you never have to take our word for it.
Tap "Verify this attestation yourself" and check the enclave against our on-chain registry, in the open.
So far: 397K private inferences, 1.7B+ tokens, 0 prompts seen by anyone.
We don't say trust us. We say check it.
6/
Tap "Verify this attestation yourself" and check the enclave against our on-chain registry, in the open.
So far: 397K private inferences, 1.7B+ tokens, 0 prompts seen by anyone.
We don't say trust us. We say check it.
6/
Then the answer proves itself. They are signed inside the enclave and verified by your browser before it reaches your screen.
The badge reads:
→ Authentic: the model you chose, run in a real TEE
→ Untampered: nothing changed on the way back
→ Private: OHTTP, encrypted
The badge reads:
→ Authentic: the model you chose, run in a real TEE
→ Untampered: nothing changed on the way back
→ Private: OHTTP, encrypted
The relay holds only ciphertext. The enclave never learns your IP.
As long as they don't collude, nothing links you to what you said.
Privacy from the wiring, not the fine print.
4/
As long as they don't collude, nothing links you to what you said.
Privacy from the wiring, not the fine print.
4/
You type your most sensitive questions into AI:
Your health, your money, the things you'd never say out loud.
Most private AI apps show you their privacy policy. That's a promise you can't audit, because a promise isn't proof.
2/
Your health, your money, the things you'd never say out loud.
Most private AI apps show you their privacy policy. That's a promise you can't audit, because a promise isn't proof.
2/
For every AI response, you can verify the security details.
→ Your device: encrypted before it leaves the browser
→ The relay: sees your IP, never your words
→ The enclave (TEE): sees your words, never your IP
→ The model: answers a request that was never tied to you
No
→ Your device: encrypted before it leaves the browser
→ The relay: sees your IP, never your words
→ The enclave (TEE): sees your words, never your IP
→ The model: answers a request that was never tied to you
No
Every AI app asks you to trust it. Ours lets you check it.
OpenGradient Chat is the private AI chat you can cryptographically verify, live, that no one read what you typed. Not even us.
Here's the tab that proves it.
(1/n)
OpenGradient Chat is the private AI chat you can cryptographically verify, live, that no one read what you typed. Not even us.
Here's the tab that proves it.
(1/n)
Here's the prompt for the video above:
A female USA gymnast performs a flawless pommel horse routine in a packed Olympic gymnastics arena. She wears a red, white, and blue Team USA leotard with subtle star detailing and "USA" lettering, hair tied back tight, chalk dust on her
A female USA gymnast performs a flawless pommel horse routine in a packed Olympic gymnastics arena. She wears a red, white, and blue Team USA leotard with subtle star detailing and "USA" lettering, hair tied back tight, chalk dust on her
Select the Quality and Filter
For the video I made, I picked Best and Standard options.
And our platform chose Seedance 2.0 for this task
- if you pick uncensored/zero filters you can try out your imaginations without limit 😅
For the video I made, I picked Best and Standard options.
And our platform chose Seedance 2.0 for this task
- if you pick uncensored/zero filters you can try out your imaginations without limit 😅
Connectez-vous pour découvrir plus de contenu