正在加载视频...

视频加载失败

返回首页

Seedance 2 just dropped as open-source.!.. ok, mostly... Helios - 14B model for minute-scale vid gen. - high-quality, fast -No KV-cache, no quantization, no TinyVAE, just pure architectural efficiency. - T2V/I2V/V2V ready.

wildmindai's profile picture

Wildminder

11,620 subscribers

25,234 次观看 • 7 个月前 •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

12 条评论

Julie Loves Tech 的头像
Julie Loves Tech7 个月前

"no KV-cache, no quantization, no TinyVAE, just pure architectural efficiency" is the flex that tells you the team was doing something genuinely different under the hood. minute-scale video generation at 14B without the usual shortcuts is a real engineering achievement.

kayinsho 的头像
kayinsho7 个月前

The big AI players are lying to everyone. They don’t need bigger LLMs or video models for better quality — they need bigger models to create a moat through cost, data control, and infrastructure barriers.

Heba AI 的头像
Heba AI7 个月前

Another repo for A100 owners only

Maurice Bourdon 的头像
Maurice Bourdon7 个月前

H100 80GB, T2V (distilled): solid temporal coherence on long clips + very fast inference: 240 frames in 2 min, 960 frames in 3 min.

Synth Seduction 的头像
Synth Seduction7 个月前

Wake me up when wan 2.5 is open source :(

The AI Optimist 的头像
The AI Optimist7 个月前

@HeyGenLabs summarize

selv Marū Atma-Goma 的头像
selv Marū Atma-Goma7 个月前

does it work with apple silicon?

Tradermatch 的头像
Tradermatch7 个月前

Interesting

Sanky 的头像
Sanky7 个月前

Minute-scale video generation open-source? Helios 14B is a game changer 🔥

SEAWP 的头像
SEAWP7 个月前

19.5FPS on a single H100 is wild

LinAi Dreams 的头像
LinAi Dreams7 个月前

Where to test? Please

Mike Codes 的头像
Mike Codes7 个月前

no kv-cache + no quantization is a bold flex. if the minute-scale clips stay temporally consistent, that’s the real win.

相关视频

🤯Real-time video generation just got HUGE. Introducing Helios: A 14B parameter model running at 19.5 FPS on a single H100. Cheaper, Faster, and Stronger than 1.3B models, generating minute-long videos without the usual tricks (even no quantization).
0:50

🤯Real-time video generation just got HUGE. Introducing Helios: A 14B parameter model running at 19.5 FPS on a single H100. Cheaper, Faster, and Stronger than 1.3B models, generating minute-long videos without the usual tricks (even no quantization).

Bin Lin

27,761 次观看 • 7 个月前

Just dropped on HF: Supertonic TTS, a blazing fast speech model. 🤯 > RTF as low as 0.001 on RTX4090 > Runs on-device (no latency, full privacy) > 66M params + 8+ language SDKs > Browser, mobile, edge — it just works. > You can adjust inference steps, batch processing, and other parameters to match your specific needs! > Open-source. Ready to build with!
0:37

Just dropped on HF: Supertonic TTS, a blazing fast speech model. 🤯 > RTF as low as 0.001 on RTX4090 > Runs on-device (no latency, full privacy) > 66M params + 8+ language SDKs > Browser, mobile, edge — it just works. > You can adjust inference steps, batch processing, and other parameters to match your specific needs! > Open-source. Ready to build with!

steven

17,784 次观看 • 10 个月前

300 million users. Mostly young women. Almost no OSINT coverage. 🇨🇳 Xiaohongshu (Little Red Book) is one of China's largest platforms - and Bellingcat just dropped a guide to using it for open-source investigations.
0:35

300 million users. Mostly young women. Almost no OSINT coverage. 🇨🇳 Xiaohongshu (Little Red Book) is one of China's largest platforms - and Bellingcat just dropped a guide to using it for open-source investigations.

The OSINT Newsletter

16,610 次观看 • 5 个月前

The future of Klever Blockchain is now fully open. The core code is officially open source — audited, transparent, and ready for the world. No black boxes. No compromises. Just secure infrastructure built to scale Web3 globally. Developers, institutions, builders: the door is open. Explore the code →
0:35

The future of Klever Blockchain is now fully open. The core code is officially open source — audited, transparent, and ready for the world. No black boxes. No compromises. Just secure infrastructure built to scale Web3 globally. Developers, institutions, builders: the door is open. Explore the code →

Klever Blockchain

316,837 次观看 • 7 个月前

🚨 Claude Code just got REPLACED overnight… for $0 An open-source project is now running a 122B model locally on your MacBook 🤯 No API fees. No cloud. No limits. Just raw power on your own machine. 1 command install. Double-click to launch. If this works at scale… it’s game over for paid AI tools. GitHub ↓
0:34

🚨 Claude Code just got REPLACED overnight… for $0 An open-source project is now running a 122B model locally on your MacBook 🤯 No API fees. No cloud. No limits. Just raw power on your own machine. 1 command install. Double-click to launch. If this works at scale… it’s game over for paid AI tools. GitHub ↓

Shruti Codes

40,749 次观看 • 6 个月前

High Republic fans. Listen, We may never have a farewell episode of The High Republic Show... And that's ok, because we do have a video I made of my first day on the High Republic set. No cuts. No script. Just pure nonsense for one minute and thirty two seconds. #starwars
1:31

High Republic fans. Listen, We may never have a farewell episode of The High Republic Show... And that's ok, because we do have a video I made of my first day on the High Republic set. No cuts. No script. Just pure nonsense for one minute and thirty two seconds. #starwars

Krystina Arielle

12,837 次观看 • 1 年前

MiniMax H3 Just Dropped on Lovart! 🔥 The world's #2 video model just landed on Lovart. Production-quality AI video at less than half the cost. •15s cinematic output with fluid, high-fidelity motion •15multimodal references: text, image, video & audio •Consistent characters, products, scenes & styles From Ads to MV to TVC, no more choosing between quality and cost.
0:21

MiniMax H3 Just Dropped on Lovart! 🔥 The world's #2 video model just landed on Lovart. Production-quality AI video at less than half the cost. •15s cinematic output with fluid, high-fidelity motion •15multimodal references: text, image, video & audio •Consistent characters, products, scenes & styles From Ads to MV to TVC, no more choosing between quality and cost.

LovartAI

241,649 次观看 • 2 个月前

🚨 THIS IS SHEIN - WITHOUT THE FILTERS Mountains of plastic-wrapped clothes piled like garbage. Workers sitting on concrete floors sorting nonstop. No music. No glam. No “haul.” Just pure volume. This is how $3 tops happen. Not magic. Not efficiency. Just scale so big you’re never supposed to see it. Fast fashion doesn’t disappear after checkout - it gets buried out of sight. So what’s worse: the conditions… or the fact most people don’t care as long as it’s cheap?
0:42

🚨 THIS IS SHEIN - WITHOUT THE FILTERS Mountains of plastic-wrapped clothes piled like garbage. Workers sitting on concrete floors sorting nonstop. No music. No glam. No “haul.” Just pure volume. This is how $3 tops happen. Not magic. Not efficiency. Just scale so big you’re never supposed to see it. Fast fashion doesn’t disappear after checkout - it gets buried out of sight. So what’s worse: the conditions… or the fact most people don’t care as long as it’s cheap?

HustleBitch

6,022,108 次观看 • 9 个月前

Game-Changer Alert! Introducing Neuro Image Gen – The ultimate AI image generator powered by Neurolov's decentralized GPU network, delivering ultra-realistic, high-resolution, and completely free image generation! ✨ No fees. No ads. No limits. Just pure AI magic. ✨ Experience DSLR-quality AI images instantly—fueled by the power of a next-gen decentralized GPU network for unmatched speed and precision! 🔗 [ Your Vision. Our AI. Let's Create Together.
0:27

Game-Changer Alert! Introducing Neuro Image Gen – The ultimate AI image generator powered by Neurolov's decentralized GPU network, delivering ultra-realistic, high-resolution, and completely free image generation! ✨ No fees. No ads. No limits. Just pure AI magic. ✨ Experience DSLR-quality AI images instantly—fueled by the power of a next-gen decentralized GPU network for unmatched speed and precision! 🔗 [ Your Vision. Our AI. Let's Create Together.

Neurolov

29,765 次观看 • 1 年前

Just used Akash Chat from for the first time. It's actually really impressive. Open source LLM model run on permissionless Akash compute. -Super fast and responsive. -No account or sign in needed. One of the few actually useful instances of crypto + AI.
0:33

Just used Akash Chat from for the first time. It's actually really impressive. Open source LLM model run on permissionless Akash compute. -Super fast and responsive. -No account or sign in needed. One of the few actually useful instances of crypto + AI.

Jesse Eckel

33,890 次观看 • 2 年前

This is crazy. invideo just dropped Sora 2 for everyone. no watermark, no 10-second cap, full-length, high-quality videos right inside invideo, OpenAI’s first official partner. Humans vs Animals. Sora 2 realism on full display, every move and camera angle behaves like a real shoot. Here are 6 wild Sora 2 examples 👇
0:32

This is crazy. invideo just dropped Sora 2 for everyone. no watermark, no 10-second cap, full-length, high-quality videos right inside invideo, OpenAI’s first official partner. Humans vs Animals. Sora 2 realism on full display, every move and camera angle behaves like a real shoot. Here are 6 wild Sora 2 examples 👇

Angry Tom

867,467 次观看 • 1 年前

You ASKED for Optimization, And we are Delivering! New patch update drops for the Luminary Demo in a few days, that adds over 35% performance boost when you are in HIGH settings. No tricks, this video doesn't use frame gen, just pure engine, lighting and texture adjustments to get things to run way better with no visual quality loss :)
0:53

You ASKED for Optimization, And we are Delivering! New patch update drops for the Luminary Demo in a few days, that adds over 35% performance boost when you are in HIGH settings. No tricks, this video doesn't use frame gen, just pure engine, lighting and texture adjustments to get things to run way better with no visual quality loss :)

Luminary | Refractive

29,448 次观看 • 8 个月前

Your local AI just got up to 5x more memory. Same model. Same device. Nearly zero accuracy loss. QVAC SDK 0.12.0 integrates TurboQuant - Google Research's latest memory optimisation algorithm. What is TurboQuant? The KV cache is the memory your model uses to track a conversation. As context grows, it fills up fast. 32K tokens. 64K. Game over. TurboQuant compresses it up to 5x with no accuracy loss. What does it unlock for you? Your app had a 16K token ceiling? It's now 96K. On the same device. Just update the QVAC SDK to get up to 5x more efficiency. No code changes. All from one SDK. The TurboQuant integration unlocks sovereign intelligence for more people, on more devices. Learn more →
1:21

Your local AI just got up to 5x more memory. Same model. Same device. Nearly zero accuracy loss. QVAC SDK 0.12.0 integrates TurboQuant - Google Research's latest memory optimisation algorithm. What is TurboQuant? The KV cache is the memory your model uses to track a conversation. As context grows, it fills up fast. 32K tokens. 64K. Game over. TurboQuant compresses it up to 5x with no accuracy loss. What does it unlock for you? Your app had a 16K token ceiling? It's now 96K. On the same device. Just update the QVAC SDK to get up to 5x more efficiency. No code changes. All from one SDK. The TurboQuant integration unlocks sovereign intelligence for more people, on more devices. Learn more →

QVAC

15,800,188 次观看 • 4 个月前

Hot off the press 🔥 AEON vLLM Ultimate → v0.24.0. Link below ⤵️ The ultimate container for AI on NVIDIA DGX Spark — vLLM compiled from source for GB10/sm_121a. 🔓 UNLOCKED (impossible until this release): ▸ DFlash + FP8 KV cache — 2× KV capacity, draft acceptance intact, measured on-device ▸ FP8 KV on vision models — the KV-bandwidth lever is back ▸ DFlash on FlashInfer & Triton backends (was FlashAttention-only) 🛠 Carried ahead of upstream (still-open PRs, merged + validated): ▸ Triton NVFP4 KV cache → up to 3× KV capacity ▸ DFlash sliding-window drafters + prefix-cache stability for DFlash ▸ High-concurrency DFlash — clean through c=64 🧰 Fixes stock v0.24.0 doesn't have: ▸ Gemma-4 tied embeddings (ModelOpt) — stock crashes at load, ours doesn't ▸ Tool calling out of the box (stock --no-deps builds 500) ▸ Unified-memory hardened — no silent KV overcommit on the 121 GB pool 📊 One image, the whole fleet: Gemma-4-26B-A4B · Qwen3.6-27B · Qwen3.6-35B-A3B (hybrid GDN) 536 tok/s @ c16 · 472 tok/s @ c12 · 262K context · image+video · tools + reasoning parsers · full CUDA-graph capture 🐳 docker pull
0:44

Hot off the press 🔥 AEON vLLM Ultimate → v0.24.0. Link below ⤵️ The ultimate container for AI on NVIDIA DGX Spark — vLLM compiled from source for GB10/sm_121a. 🔓 UNLOCKED (impossible until this release): ▸ DFlash + FP8 KV cache — 2× KV capacity, draft acceptance intact, measured on-device ▸ FP8 KV on vision models — the KV-bandwidth lever is back ▸ DFlash on FlashInfer & Triton backends (was FlashAttention-only) 🛠 Carried ahead of upstream (still-open PRs, merged + validated): ▸ Triton NVFP4 KV cache → up to 3× KV capacity ▸ DFlash sliding-window drafters + prefix-cache stability for DFlash ▸ High-concurrency DFlash — clean through c=64 🧰 Fixes stock v0.24.0 doesn't have: ▸ Gemma-4 tied embeddings (ModelOpt) — stock crashes at load, ours doesn't ▸ Tool calling out of the box (stock --no-deps builds 500) ▸ Unified-memory hardened — no silent KV overcommit on the 121 GB pool 📊 One image, the whole fleet: Gemma-4-26B-A4B · Qwen3.6-27B · Qwen3.6-35B-A3B (hybrid GDN) 536 tok/s @ c16 · 472 tok/s @ c12 · 262K context · image+video · tools + reasoning parsers · full CUDA-graph capture 🐳 docker pull

ÆON FORGE ✨

24,800 次观看 • 3 个月前

$26k+/day ads are using formats like this boyfriend blind scent tests on camera real reactions, no overthinking, pure curiosity it feels like content, not marketing claude writes the flow and dialogue seedance turns it into realistic clips instantly no creators to coordinate no reshoots or production delays just prompt → interaction → ready to run same format reused across products different couples, reactions, outcomes every time one idea becomes dozens of videos fast that’s why these scale so easily rt + comment “scent” and i’ll send the setup (follow for dm)
0:21

$26k+/day ads are using formats like this boyfriend blind scent tests on camera real reactions, no overthinking, pure curiosity it feels like content, not marketing claude writes the flow and dialogue seedance turns it into realistic clips instantly no creators to coordinate no reshoots or production delays just prompt → interaction → ready to run same format reused across products different couples, reactions, outcomes every time one idea becomes dozens of videos fast that’s why these scale so easily rt + comment “scent” and i’ll send the setup (follow for dm)

Alessandro

16,486 次观看 • 5 个月前

Something BIG is coming to Akool… Seedance 2.0 is on the way and it’s seriously next level. We’re talking about a model so powerful, you can basically create anything you imagine. No limits. No complicated workflows. Just pure creativity. From idea → video in seconds Faster, smarter, more flexible Built for creators, marketers, and storytellers Get ready to create without boundaries. Coming soon. Stay tuned.
0:46

Something BIG is coming to Akool… Seedance 2.0 is on the way and it’s seriously next level. We’re talking about a model so powerful, you can basically create anything you imagine. No limits. No complicated workflows. Just pure creativity. From idea → video in seconds Faster, smarter, more flexible Built for creators, marketers, and storytellers Get ready to create without boundaries. Coming soon. Stay tuned.

Akool Inc

827,973 次观看 • 7 个月前

This 22-year-old American developer showed how he makes $14,200 a month using just a dual-GPU setup and Cohere’s new open-weights model. He doesn't have a powerhouse dev team. No massive VC funding. No expensive enterprise infrastructure. He built a "localized AI workflow" powered by the newly dropped Command A+ model → an ultra-efficient system he calls “algorithmic mining.” The system works like this: 1 open-weights model → Apache 2.0 license integration → Command A+'s Mixture of Experts (MoE) processes complex data at lightning speed → Quantized architecture runs smoothly on just 2 GPUs → Automated agents handle enterprise-level tasks while he sleeps. 3 recurring revenue streams from a 72-hour implementation: > $49/month Essential API access > $99/month Scale tier for small teams > $199/month Custom enterprise pipeline > Delivered with near-zero hosting latency > Total setup time: 3 days He didn't wait for massive cloud budget approval. He took a highly practical, open-source-friendly model, spun it up locally, and built the exact kind of agents usually reserved for Fortune 500 companies. As the developer put it perfectly: In the open-weight era, efficiency and raw speed are the ultimate leverage.
1:54

This 22-year-old American developer showed how he makes $14,200 a month using just a dual-GPU setup and Cohere’s new open-weights model. He doesn't have a powerhouse dev team. No massive VC funding. No expensive enterprise infrastructure. He built a "localized AI workflow" powered by the newly dropped Command A+ model → an ultra-efficient system he calls “algorithmic mining.” The system works like this: 1 open-weights model → Apache 2.0 license integration → Command A+'s Mixture of Experts (MoE) processes complex data at lightning speed → Quantized architecture runs smoothly on just 2 GPUs → Automated agents handle enterprise-level tasks while he sleeps. 3 recurring revenue streams from a 72-hour implementation: > $49/month Essential API access > $99/month Scale tier for small teams > $199/month Custom enterprise pipeline > Delivered with near-zero hosting latency > Total setup time: 3 days He didn't wait for massive cloud budget approval. He took a highly practical, open-source-friendly model, spun it up locally, and built the exact kind of agents usually reserved for Fortune 500 companies. As the developer put it perfectly: In the open-weight era, efficiency and raw speed are the ultimate leverage.

Ridark

214,085 次观看 • 4 个月前

We’re not generating images anymore… We’re generating scenes 🎬 This started as a single frame made in GPT Image 2. Then I dropped it into Seedance 2.0 via Pollo AI… And it turned into a full moving shot: Camera motion Lighting shifts Real depth No editing No timeline No VFX team Just image → motion Built for brands, ads, and content teams that want to turn ideas into production-ready videos Faster and at lower cost This is what creation looks like now. Follow for more.
1:49

We’re not generating images anymore… We’re generating scenes 🎬 This started as a single frame made in GPT Image 2. Then I dropped it into Seedance 2.0 via Pollo AI… And it turned into a full moving shot: Camera motion Lighting shifts Real depth No editing No timeline No VFX team Just image → motion Built for brands, ads, and content teams that want to turn ideas into production-ready videos Faster and at lower cost This is what creation looks like now. Follow for more.

marium

46,700 次观看 • 5 个月前