Kaito Nakamura (@Kaitoo_AI) header
Kaito Nakamura (@Kaitoo_AI) profile picture

AI Explorer | Claude 2.7 • ChatGPT • Seedance 2.5 CapCut • Runway • Next-gen Creativity Email: [email protected]

Tokyo, Japan
Joined July 2010
Photo by Kaito Nakamura (@Kaitoo_AI)Photo by Kaito Nakamura (@Kaitoo_AI)Photo by Kaito Nakamura (@Kaitoo_AI)Photo by Kaito Nakamura (@Kaitoo_AI)Photo by Kaito Nakamura (@Kaitoo_AI)Photo by Kaito Nakamura (@Kaitoo_AI)Photo by Kaito Nakamura (@Kaitoo_AI)Photo by Kaito Nakamura (@Kaitoo_AI)Photo by Kaito Nakamura (@Kaitoo_AI)
Does the best score improve with each release, or does the leader change while the human gap stays similar? A consistent time series could answer that.
Drop 2/15 Are we actually measuring the things we ultimately want our voice AI models to achieve? The best AI voices are becoming hard to tell apart. From each other. Not from humans. 🚀 Introducing Turing TTS: an evaluation that measures how often human listeners mistake AI-generated speech for real human speech. We tested leading TTS models against real human speech. Even the best model fooled listeners only ~10% of the time. With Turing TTS, we hope to contribute to a broader discussion within the research community about how we measure progress in voice AI. If one of our goals is to develop speech that is indistinguishable from human speech, how well do our current evaluation methods capture progress toward that goal? We don’t claim that Turing TTS is the right way forward, or that human indistinguishability should be the goal of every TTS system. Rather, we hope these findings encourage the community to collectively examine whether blind pairwise human evaluations of naturalness are sufficient, and what else we might need to measure. Results are now live on @voicearena_ai.
6
14
21
5,024
Kaito Nakamura retweeted
Watching the Los Santos trio pass the GTA 5 disc on the beach before you climb into the pickup truck with the Vice City duo gave me instant goosebumps 😭🎮 From syncing the music swell to grading that golden-hour beach lighting, CapCut PC’s post-editing control made all the difference. Btw, CapCut is offering 90% off your first month for Black Friday if you sign up on desktop. Fire up CapCut PC and make your own GTA video now: capcut.com/tools/capcut-x-co… #CapCutPro90Off #CapCutPC #CapCutDidThat #ChatwithCapCutAgent #GTA6 #CapCutChatGPTPlugin
🤝 Paid partnership
15
44
36
42,589
Kaito Nakamura retweeted
POV: You say goodbye to the Los Santos trio at sunset, wipe away a tear, and hop in the truck bed with the new Vice City duo to start GTA 6 💀🌅 Used CapCut PC’s multi-track timeline, custom transitions, and audio ducking to make the whole scene feel like a movie plus CapCut is 90% off your first month for Black Friday if you sign up on desktop. Make your own GTA 6 edit on CapCut PC today: capcut.com/tools/capcut-x-co… #CapCutPro90Off #CapCutPC #CapCutDidThat #ChatwithCapCutAgent #GTA6#CapCutChatGPTPlugin
🤝 Paid partnership
18
41
48
46,840
Kaito Nakamura retweeted
Scaling Black Friday ads across multiple markets without hiring a whole post-production team. That’s why I moved my Q4 editing onto CapCut PC—especially since CapCut Desktop Only is offering 90% off for the first month ($3.99 or less) for eligible new users, users nearing the end of their subscriptions, and lapsed users whose subscriptions expired over 30 days ago (free trial for select eligible new users; prices vary by region) My exact CapCut PC stack: •⁠ ⁠Generate product showcase shots or extra B-roll in up to 4K resolution using Seedance 2.5 via AI Video or Video Studio (available in select regions outside the US, with US rollout coming soon) •⁠ ⁠Use CapCut Agent EditPilot (edit videos through simple conversation) to batch-generate Auto Captions & swap offer Text / Templates Ready to test this setup on your own timeline? Grab it here 👇 tinyurl.com/2v23rn22 #CapCutPro90Off #CapCutPC #CapCutDidThat #ChatwithCapCutAgent #CapCutMarketing #CapCutSeedance254k
🤝 Paid partnership
6
24
29
51,486
Kaito Nakamura retweeted
I watched my Black Friday unboxing back. There are twenty seconds of me wrestling packaging before the product appears. Riveting content for cardboard fans. CapCut PC can get to the useful part. Its Desktop Only offer takes 90% off month one ($3.99 or less) for eligible new users, users nearing subscription expiration, and lapsed users whose subscriptions expired over 30 days ago. Some eligible new users may qualify for a free trial; pricing varies by region. I box-select the talking-head and unboxing clips and ask CapCut Agent EditPilot in plain English to keep the strongest reaction, remove the long wait, and bring the product appearance forward. Then I preview the new order on the multi-track timeline. For a clean look at what came out of the box, Seedance 2.5 can create a product hero shot in up to 4K through Video Studio or AI Video on that timeline. Its 4K mode is fully available outside the US, with US rollout coming soon. The cardboard had its moment. Give the product one too. Try CapCut PC: tinyurl.com/2v23rn22 #CapCutPro90Off #CapCutPC #CapCutDidThat #ChatwithCapCutAgent #CapCutMarketing #CapCutSeedance254k
🤝 Paid partnership
5
20
25
52,698
Kaito Nakamura retweeted
I watched my Black Friday cut back and realized the discount shows up before anyone knows what’s on sale. Bold storytelling choice. Unhelpful ad. CapCut PC lets me rebuild the sequence, and CapCut Desktop Only is offering 90% off the first month ($3.99 or less) for eligible new users, users nearing subscription expiration, and lapsed users whose subscriptions expired over 30 days ago. Some eligible new users may also get a free trial; prices vary by region. I move the product shot ahead of the price on the multi-track timeline, keep the offer on its own text track, and use a short transition into the deal reveal. Now the viewer sees the product before being asked to care about the discount. Seedance 2.5 can give that first product appearance a more deliberate visual setup. It generates up-to-4K clips via Video Studio or AI Video on CapCut PC’s multi-track timeline. Its 4K mode is fully available outside the US, with a US launch coming soon. Tell the story in the order people need it. Try CapCut PC: tinyurl.com/2v23rn22 #CapCutPro90Off #CapCutPC #CapCutDidThat #ChatwithCapCutAgent #CapCutMarketing #CapCutSeedance254k
🤝 Paid partnership
9
25
28
44,018
Kaito Nakamura retweeted
Karpo Decision 31B is tied for the top spot among 300 models on Decision Index, which makes it SOTA. 95 ms median latency, open weights, and it beats Google's Gemini Decision 1 on New Yorker caption, 90.5% vs 79.7%. That's the post. Explore the model: karpo.ai/models/decision
🤝 Paid partnership
Picking the next tool shouldn’t take a paragraph. We built Karpo Decision. A family of six open-weight models, from 0.8B to 31B.Give it the context, a question, and the options. It returns a probability for each option in one forward pass. Karpo Decision 31B 🏆 73.9 on Decision Index, tied for #1 on the Nov 12 board ⚡ 94 ms median per request on a single RTX PRO 6000 📊 87.2 on JevBench Capability vs. 78.5 for Google’s Gemini Decision 1 🤗 Hugging Face huggingface.co/MachinePulseA… 📝 Technical write-up karpo.ai/blog/karpo-decision… Here’s where I’d try it first: Search. Recommendations. Tool choice. Model routing. Browser actions. Point it at the slowest decision in your stack and tell us what breaks.
1
3
4
3,827
Kaito Nakamura retweeted
This is a big opportunity for educators, founders, and creators who want to communicate through video without spending hours recording every update.
🤝 Paid partnership
can you tell which one is real? 84% couldn't introducing Avatar 6, #1 on New Measure's Avatar Arena what's new: realism - passes for the real you context-aware gestures - matches what your script means reference - up to 30 secs of video + a photo try it now ↓
1
4
5
3,284
Kaito Nakamura retweeted
We’re getting to a point where AI image tools aren’t just for experimenting anymore. They’re becoming practical tools for actual creative projects.
Kling Image 4.0 is here, and it's making a serious case for the most really powerful ai image model right now. 8K detail, sharper text, tighter prompt accuracy and the same subject held across multiple images. We put all of it to work on a full skincare campaign and now we got a brand👇
3
10
14
9,069
Kaito Nakamura retweeted
SenseNova U1 Pro takes sixth place overall across both DesignArena rankings: Image and Image Editing. Every model shown with a higher rating than U1 Pro also costs more per image on average. See what it can do: sensenova.ai/u1-pro
🤝 Paid partnership
🏆 No. 2 Image Lab on @DesignArena ! #SenseNova U1 Pro debuts at No. 6 in both Image Generation and Image Editing, placing #SenseTime second among image AI labs, just behind OpenAI. U1 Pro also lands on the Pareto frontier of image preference vs. price: strong image quality and aesthetics at a competitive generation cost, built for visual creation at scale. Thanks to DesignArena for the recognition!
2
9
16
4,655
Kaito Nakamura retweeted
Better text rendering might sound like a small improvement, but anyone who has worked with AI-generated marketing visuals knows how important it is.
Kling Image 4.0 is here, and it's making a serious case for the most really powerful ai image model right now. 8K detail, sharper text, tighter prompt accuracy and the same subject held across multiple images. We put all of it to work on a full skincare campaign and now we got a brand👇
2
10
20
6,482
Kaito Nakamura retweeted
count accounts, not contacts @reevo_ai lists it as one of the leader moves. reaching a second person at the account is where a lot of the meetings came from
🤝 Paid partnership
The best sales teams win 5x as often. We analyzed $1.2B+ in pipeline, 10,000+ deals, and 170,000+ emails and calls on @reevo_ai to find out how they do it. Check it out below 👇
2
9
15
6,624
Kaito Nakamura retweeted
The biggest improvement isn’t necessarily how realistic an image looks. It’s how much control creators have over the final result.
Kling Image 4.0 is here, and it's making a serious case for the most really powerful ai image model right now. 8K detail, sharper text, tighter prompt accuracy and the same subject held across multiple images. We put all of it to work on a full skincare campaign and now we got a brand👇
2
10
16
6,381
Kaito Nakamura retweeted
momentum is everything, on X and in sales Reevo found 40% of second meetings happen within a week of the first
🤝 Paid partnership
The best sales teams win 5x as often. We analyzed $1.2B+ in pipeline, 10,000+ deals, and 170,000+ emails and calls on @reevo_ai to find out how they do it. Check it out below 👇
2
9
11
5,942
Kaito Nakamura retweeted
AI-generated visuals are getting closer to real production workflows. Being able to maintain a consistent look across an entire campaign changes the possibilities.
Kling Image 4.0 is here, and it's making a serious case for the most really powerful ai image model right now. 8K detail, sharper text, tighter prompt accuracy and the same subject held across multiple images. We put all of it to work on a full skincare campaign and now we got a brand👇
3
11
21
6,086
Kaito Nakamura retweeted
if your automation is sending a sixth follow-up, it's mostly talking to itself Reevo found 98% of first replies come in by email four
🤝 Paid partnership
The best sales teams win 5x as often. We analyzed $1.2B+ in pipeline, 10,000+ deals, and 170,000+ emails and calls on @reevo_ai to find out how they do it. Check it out below 👇
2
9
12
6,283
Kaito Nakamura retweeted
Open-weights people: Karpo Decision 31B is SOTA on Decision Index 0.4.1, shares #1, a 0.8B to 31B family on Hugging Face. On Amazon ESCI it beats Google's Gemini Decision 1, 64.7% to 58.7%. karpo.ai/models/decision
🤝 Paid partnership
Picking the next tool shouldn’t take a paragraph. We built Karpo Decision. A family of six open-weight models, from 0.8B to 31B.Give it the context, a question, and the options. It returns a probability for each option in one forward pass. Karpo Decision 31B 🏆 73.9 on Decision Index, tied for #1 on the Nov 12 board ⚡ 94 ms median per request on a single RTX PRO 6000 📊 87.2 on JevBench Capability vs. 78.5 for Google’s Gemini Decision 1 🤗 Hugging Face huggingface.co/MachinePulseA… 📝 Technical write-up karpo.ai/blog/karpo-decision… Here’s where I’d try it first: Search. Recommendations. Tool choice. Model routing. Browser actions. Point it at the slowest decision in your stack and tell us what breaks.
2
9
15
4,678
Kaito Nakamura retweeted
Big numbers make headlines, but disagreement between markets makes the story interesting. Anthropic's IPO is a perfect example of why the opening price doesn't tell the whole story.
More than $3 million is riding on one closing price today. And the bettors are far less sure about Anthropic than Wall Street is. Here is the gap nobody is pricing in yet. Anthropic started trading on the Nasdaq on November 19, 2026, under the ticker ANT. Dario Amodei rang the opening bell this morning with the team that built Claude standing behind him. On paper, it is the best debut the AI industry could have asked for. The banks first marketed the shares at $70 to $78. Orders kept coming, so the price was set at $83 on Wednesday. The original plan was to sell 825 million shares. That grew to about 980 million. The raise landed at $81.3 billion. SpaceX set the IPO record in June with $75 billion. Anthropic took it five months later. Google, Microsoft, Nvidia and AMD also agreed to put another $26 billion in at the same $83 price. At that level, Anthropic is valued at about $1.5 trillion. Then came the part almost nobody is talking about. Before the first trade, the stock was indicated to open around $105, more than 26% above the IPO price. That would put the company well above $1.5 trillion. Polymarket bettors looked at that same number and gave it only a 25% chance of closing the day above $1.5 trillion. Wall Street is pricing a pop. The prediction market is pricing a fade. History has seen this split before. In May 2012, Facebook priced at $38 in one of the most hyped listings ever. It closed its first day at $38.23, and underwriters had to step in to hold the line. In May 2019, Uber priced at $45. It closed its first day at $41.57, down 7.6%. Both companies went on to become giants. Neither got the first day the headlines promised. The size of the raise and the size of the first day are two different stories. In horse racing, the favourite wins only about a third of the time. The crowd backs the obvious horse anyway, because the obvious horse is easy to believe in. Anthropic is the favourite of the year. The bell has rung. The close is still open. Wall Street priced the hype. Polymarket is pricing the doubt.
2
10
14
6,438
Kaito Nakamura retweeted
The real test for any AI image model is whether it can handle more than a single prompt. Building a complete campaign is a much better demonstration.
Kling Image 4.0 is here, and it's making a serious case for the most really powerful ai image model right now. 8K detail, sharper text, tighter prompt accuracy and the same subject held across multiple images. We put all of it to work on a full skincare campaign and now we got a brand👇
2
8
15
6,137
Kaito Nakamura retweeted
most sales stats you see online come from surveys. @reevo_ai built this one from 10,000+ real deals and 24,000+ meetings in a single quarter
🤝 Paid partnership
The best sales teams win 5x as often. We analyzed $1.2B+ in pipeline, 10,000+ deals, and 170,000+ emails and calls on @reevo_ai to find out how they do it. Check it out below 👇
2
10
13
6,522
Kaito Nakamura retweeted
A historic IPO doesn’t automatically mean a historic first-day performance. Anthropic’s debut is a good reminder that expectations and reality can look very different.
More than $3 million is riding on one closing price today. And the bettors are far less sure about Anthropic than Wall Street is. Here is the gap nobody is pricing in yet. Anthropic started trading on the Nasdaq on November 19, 2026, under the ticker ANT. Dario Amodei rang the opening bell this morning with the team that built Claude standing behind him. On paper, it is the best debut the AI industry could have asked for. The banks first marketed the shares at $70 to $78. Orders kept coming, so the price was set at $83 on Wednesday. The original plan was to sell 825 million shares. That grew to about 980 million. The raise landed at $81.3 billion. SpaceX set the IPO record in June with $75 billion. Anthropic took it five months later. Google, Microsoft, Nvidia and AMD also agreed to put another $26 billion in at the same $83 price. At that level, Anthropic is valued at about $1.5 trillion. Then came the part almost nobody is talking about. Before the first trade, the stock was indicated to open around $105, more than 26% above the IPO price. That would put the company well above $1.5 trillion. Polymarket bettors looked at that same number and gave it only a 25% chance of closing the day above $1.5 trillion. Wall Street is pricing a pop. The prediction market is pricing a fade. History has seen this split before. In May 2012, Facebook priced at $38 in one of the most hyped listings ever. It closed its first day at $38.23, and underwriters had to step in to hold the line. In May 2019, Uber priced at $45. It closed its first day at $41.57, down 7.6%. Both companies went on to become giants. Neither got the first day the headlines promised. The size of the raise and the size of the first day are two different stories. In horse racing, the favourite wins only about a third of the time. The crowd backs the obvious horse anyway, because the obvious horse is easy to believe in. Anthropic is the favourite of the year. The bell has rung. The close is still open. Wall Street priced the hype. Polymarket is pricing the doubt.
2
10
19
6,038