Ashley Reese | AI

[Ashley Reese | AI @ASHLEYREESE_AI]

third mirror link in this thread now. Wan 3.0's official home is Qwen Create at https://t.co/mXj9mnsazw — don't put a card into anything else.

[RAVI KUMAR SAHU @RAVIKUMARSAHU78]

WAN 3.0 is seriously impressive.

The video quality, references, and control make it feel like a real step forward.

And the best part? You can access WAN 3.0 on the Pika API Club for less expensive than other platforms.

Making powerful video models more accessible is exactly what creators need.

[RAVI KUMAR SAHU @RAVIKUMARSAHU78]

WAN 3.0 is seriously impressive.

The video quality, references, and control make it feel like a real step forward.

And the best part? You can access WAN 3.0 on the Pika API Club for less expensive than other platforms.

Making powerful video models more accessible is exactly what creators need.

[Ashley Reese | AI @ASHLEYREESE_AI]

@Parul_Gautam7 i've run out of reasons to be smug about Wan 3.0, which i've been running on https://t.co/mXj9mnsazw all week.

[Parul Gautam @Parul_Gautam7]

I’ve been testing a lot of AI video tools lately, and one thing still breaks pretty quickly: longer scenes.

A clip might look great for a few seconds, then the character, camera or audio starts drifting.

Wan 3.0 on Magnific feels different.

https://t.co/LbOx3LRG8X

[Ashley Reese | AI @ASHLEYREESE_AI]

right, since everyone's asking in the DMs.

no, i don't think this ends filmmaking. i've been hearing that since digital, since DSLRs, since phones, since drones. every time the same panic, every time the work got broader and stranger and mostly better.

what it does is remove the cost of finding out whether an idea works. that's the whole thing. i used to get one shot at a concept because it cost money to see it. this week i got to version nine.

version nine was better than version one. obviously it was. that's how anything gets made.

it's Wan 3.0. i've been using it through Qwen Create. if you make things, go and use it. if you're here to be angry about it that's fine too, but you're going to be angry for a long time.

[Parul Gautam @Parul_Gautam7]

I’ve been testing a lot of AI video tools lately, and one thing still breaks pretty quickly: longer scenes.

A clip might look great for a few seconds, then the character, camera or audio starts drifting.

Wan 3.0 on Magnific feels different.

https://t.co/LbOx3LRG8X

[Skild AI @SkildAI]

Introducing S1, our new foundation model that learns from one example.

It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning.

Watch S1 operate in real-time via in-context learning:

[Ashley Reese | AI @ASHLEYREESE_AI]

@Jubayer7792821 Game Changer

[Jubayer @Jubayer7792821]

💥JUST IN: Seedance 2.5 and Minimax-H3 are now officially live on GlobalGPT!

Generate stable 30 sec videos in one click. Seedance 2.5 can cut your marketing production costs by 10x.

Smart teams are already using it to create ads, short dramas, IP characters, and more👇

[Ashley Reese | AI @ASHLEYREESE_AI]

@Mohibur_Tech_Ai Game Changer

[Error Mohibur @Mohibur_Tech_Ai]

Building an AI agent usually sounds complicated, but AITOPIA is making it much simpler.

You can create one without coding, publish it on their marketplace, and earn 70% revenue share.

If you’ve got an idea for an AI agent, this seems like a good time to try it out. 🚀

[Ashley Reese | AI @ASHLEYREESE_AI]

@kiranjutt1714 Sound Good

[Ashley Reese | AI @ASHLEYREESE_AI]

@Honeybee_0ee Awesome

[Ashley Reese | AI @ASHLEYREESE_AI]

@drinayakhan Sound Good

[Ashley Reese | AI @ASHLEYREESE_AI]

The next generation of AI is not just about answering.

It is about adapting.

@dotsstudioai’s dots3-note preview explores how AI can work on complex tasks over time.

Powered by:

280B total parameters

16B active parameters

512K context window

it combines three key capabilities.

🧠 TEMPO

Helps agents reflect during execution, evaluate progress, and adjust decisions instead of blindly following a plan.

🌱 Learning in deployment

Allows AI to explore unfamiliar environments, form hypotheses, and update memory.

🌍 Multimodal understanding

Lets AI combine text, vision, and audio to better understand reality.

Together, these capabilities create a more adaptive AI agent.

This is the direction behind rednote and the same level of intelligence demonstrated by achievements like IMO gold medal-level reasoning.

AI is moving from static answers toward systems that can learn, reason, and adapt.

Explore: https://t.co/Cl5xdfdzVP

Api free now: https://t.co/YjsmYSlJnN

[Ashley Reese | AI @ASHLEYREESE_AI]

The biggest savings usually appear after the first few generations.
When you create daily, every cent per second compounds.
Dreamina’s Seedance 2.0 Fast starts at $0.026/sec.

[Manish Kumar Shah @manishkumar_dev]

If you’re creating AI videos, the pricing of the model matters too 👀

Most people are paying Seedance 2.5 prices for work 2.0 already handles.

Seedance 2.0 & 2.0 Fast — cheapest on Dreamina, from $0.026/sec.

Dreamina Seedance 2.0 gives creators access to Seedance 2.0 + Seedance 2.0 fast at significantly lower generation costs.

The Dreamina platform offers:

• Seedance 2.0 720p from $0.055/sec
• Seedance 2.0 fast 720p from $0.026/sec
• Annual Advanced plan at $335
• Save up to 76% per second compared with other leading platforms

For creators who generate AI videos regularly, this can make a big difference in the long run.

#Dreamina #DreaminaPartner

[Manish Kumar Shah @manishkumar_dev]

If you’re creating AI videos, the pricing of the model matters too 👀

Most people are paying Seedance 2.5 prices for work 2.0 already handles.

Seedance 2.0 & 2.0 Fast — cheapest on Dreamina, from $0.026/sec.

Dreamina Seedance 2.0 gives creators access to Seedance 2.0 + Seedance 2.0 fast at significantly lower generation costs.

The Dreamina platform offers:

• Seedance 2.0 720p from $0.055/sec
• Seedance 2.0 fast 720p from $0.026/sec
• Annual Advanced plan at $335
• Save up to 76% per second compared with other leading platforms

For creators who generate AI videos regularly, this can make a big difference in the long run.

#Dreamina #DreaminaPartner

[Ashley Reese | AI @ASHLEYREESE_AI]

I decided to test Omneky with the simplest input possible: just my website.

It used that to generate an ad powered by Seedance 2.5, and I honestly didn’t expect the result to look this polished.

The visuals were clean, the ad actually made sense for my website, and the whole process was surprisingly simple.

I’ve tried plenty of AI tools, but getting a result like this from just a website genuinely caught me off guard. Pretty impressive. 👀

[Ashley Reese | AI @ASHLEYREESE_AI]

@helio Try it → https://t.co/yQeXI3bucf
Come ask us things → https://t.co/JiHSrSDHNB

[Ashley Reese | AI @ASHLEYREESE_AI]

I built a complete index of the open-source harness projects submitted to DeepSeek's beta-tester call with @helio Half a day of comment-section digging, delivered as a published report.

Twitter authorization, GitHub inspection, 618 repos read, categories assigned, front-end design, deploy I did none of it. I described the outcome once, and Helio owned the path.

This run happened to be open-source projects in a DeepSeek thread. The same Helio run works on complaints under a competitor's launch, or new tools appearing in your space.

Any job that starts with "go read the comments" can be handed off now.

Full report → https://t.co/Tj2kOMk4hs

[Ashley Reese | AI @ASHLEYREESE_AI]

@HeyGen you controlling the smoothing level is the detail that matters