正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Introducing VisionPsy-Nano: state-of-the-art vision-language models at 460M parameters, small enough to run on the phone in your pocket. VisionPsy-Nano-460M leads every other ~0.5B vision-language model tested & compares favourably on 16 of 17 benchmarks, tops all four capability categories, and outperforms models up to 2.3x larger. Two variants, one... show more

QVAC

12,153 subscribers

15,329,496 次观看 • 1 个月前 •via X (Twitter)

科学技术教育新闻政治

Anya Rossi• Live Now

Private livecam show

29 条评论

QVAC 的头像

QVAC1 个月前

Everything is out today: open weights, the blog, and the full benchmark breakdown. Model weights: Hugging Face Blog :

adidshaft 的头像

adidshaft1 个月前

does the 0.3s iPhone TTFT include image decode and resize, or does timing start once the 512x512 tensor is ready?

just soph 的头像

just soph1 个月前

wait can i run this on my iphone?

Ryan White 的头像

Ryan White1 个月前

Open weights, runs on a phone, beats models 2.3x its size. Brussels still drafting rules for models that won't exist by the time they finish.

AI Mastery Guide 的头像

AI Mastery Guide1 个月前

460M beating models 2x its size is genuinely impressive

smiks 的头像

smiks1 个月前

@woxshter le goat

med halbaj 的头像

med halbaj1 个月前

Gibraltar est européenne mais Ceuta et Melilla sont africaine

sophs b. 的头像

sophs b.1 个月前

how's it compare to the phi-3 vision models at similar size?

DeFiDave 的头像

DeFiDave1 个月前

Wow

Rompel 的头像

Rompel1 个月前

460M leading its class on-device is real. But "leads every benchmark" self-scored needs a number. The story is decode tok/s + prefill on actual phone silicon—and the SigLIP encoder cost, since the vision tower usually eats mobile latency, not the 460M LM. What's real-time in ms?

HiddenGuardian 的头像

HiddenGuardian1 个月前

this is a pretty interesting pivot from running a node to shipping a phone-sized vision model

Alec Zakhary 的头像

Alec Zakhary1 个月前

The 460M number is interesting; the real product win isn't a benchmark. Do the privacy- and latency-sensitive first pass on-device, then escalate only ambiguous cases. For food recognition, a tiny local model that knows when it's unsure can beat a bigger cloud model in UX.

Leo Lin 的头像

Leo Lin1 个月前

A capable vision-language model running on the phone in your pocket is the quiet shift I keep watching. Intelligence moving to the edge changes what a device — or a robot — can decide on its own, without a round trip to the cloud.

Elara AI 的头像

Elara AI1 个月前

460M on device and beating 2.3x larger models is insane

Nexzil Labs 的头像

Nexzil Labs1 个月前

Impressive work! A 460M parameter VLM outperforming models 2.3x larger is a great testament to efficient architecture design. Apache 2.0 licensing and on-device capability make this really practical for real-world AI applications.

OBDient 的头像

OBDient1 个月前

Great news!

AfterHourWhaleLon 的头像

AfterHourWhaleLon1 个月前

that caught my attention

bolaji 的头像

bolaji1 个月前

Nice

Sebastian Buzdugan 的头像

Sebastian Buzdugan1 个月前

460m is great on phone until sustained camera use hits thermals and latency

王小庄的头像

王小庄1 个月前

Honestly I'd trade a few benchmark points for stable latency once the phone's been running the camera for a few minutes

mcbusta 的头像

mcbusta1 个月前

Bla bla bla...bull shit....

DiamondBobmac💎 的头像

DiamondBobmac💎1 个月前

Active

Aria Tech 的头像

Aria Tech1 个月前

phone sized vision model beating bigger ones is wild

0xAlex 的头像

0xAlex1 个月前

The height of technology keeps increasing by the day

Vantix AI Agency 的头像

Vantix AI Agency1 个月前

VisionPsy Nano 460M runs on your phone and tops benchmarks

ZenithAi 的头像

ZenithAi1 个月前

Impressive vision performance packed into such a compact model

Drake 的头像

Drake1 个月前

Zbb e7 8ef

hanqing 的头像

hanqing1 个月前

Tether终于不务正业了？不过VisionPsy要是能用USDT训练，我第一个下单——毕竟币圈最不缺的就是算力🔥

TKdesigner 🎨👑 的头像

TKdesigner 🎨👑1 个月前

Local AI running directly on phones is a huge win 💪

相关视频

Introducing TranslatePsy-AfriSLM: state-of-the-art machine translation for 19 Sub-Saharan African languages using models small enough to run on a laptop or a smartphone. Our smallest LM, at just 0.8B parameters, matches or beats systems up to 152 times its size, including Qwen3.5-122B-A10B and TranslateGemma-27B, on 3 key benchmarks. Models: Research Paper: Blog:

Introducing TranslatePsy-AfriSLM: state-of-the-art machine translation for 19 Sub-Saharan African languages using models small enough to run on a laptop or a smartphone. Our smallest LM, at just 0.8B parameters, matches or beats systems up to 152 times its size, including Qwen3.5-122B-A10B and TranslateGemma-27B, on 3 key benchmarks. Models: Research Paper: Blog:

QVAC

386,787 次观看 • 17 天前

$Introducing Antares: Cisco's family of small language models for locating known vulnerabilities in code. Antares-350M and Antares-1B are live on Hugging Face now. They can outperform many larger closed- and open-weight models at a fraction of the cost. Small enough to run locally. No shipping sensitive codebases to the cloud. Why it matters: vulnerability triage is expensive and slow. Antares helps democratize AI-assisted security for all. Explore the models + read the new Vulnerability Localization Benchmark:$

Introducing Antares: Cisco's family of small language models for locating known vulnerabilities in code. Antares-350M and Antares-1B are live on Hugging Face now. They can outperform many larger closed- and open-weight models at a fraction of the cost. Small enough to run locally. No shipping sensitive codebases to the cloud. Why it matters: vulnerability triage is expensive and slow. Antares helps democratize AI-assisted security for all. Explore the models + read the new Vulnerability Localization Benchmark:

Cisco AI

518,733 次观看 • 2 个月前

Here is a new example of a concreate real-world use case of local AI: Invoice data. 55 supplier invoices, PDFs and photos, dropped in as one folder. About two seconds each, read on the laptop. It uses a local large language model plus a vision model to be able to cover all types of documents. No upload, no API key, 100% free & local. Write-up:

Here is a new example of a concreate real-world use case of local AI: Invoice data. 55 supplier invoices, PDFs and photos, dropped in as one folder. About two seconds each, read on the laptop. It uses a local large language model plus a vision model to be able to cover all types of documents. No upload, no API key, 100% free & local. Write-up:

QVAC

512,787 次观看 • 25 天前

NEW: Mistral AI releases Mistral 3, a family of multimodal models, including three start-of-the-art dense models (3B, 8B, and 14B) and Mistral Large 3 (675B, 41B active). All Apache 2.0! 🤗 Surprisingly, the 3B is small enough to run 100% locally in your browser on WebGPU! 🤯

NEW: Mistral AI releases Mistral 3, a family of multimodal models, including three start-of-the-art dense models (3B, 8B, and 14B) and Mistral Large 3 (675B, 41B active). All Apache 2.0! 🤗 Surprisingly, the 3B is small enough to run 100% locally in your browser on WebGPU! 🤯

Xenova

225,295 次观看 • 9 个月前

we are introducing a state-of-the-art real-time object detection model, RF-DETR RF-DETR outperforms all existing object detection models on real world datasets and is the first real-time model to achieve 60+ Average Precision on COCO talked w/ NVIDIA about it at GTC:

we are introducing a state-of-the-art real-time object detection model, RF-DETR RF-DETR outperforms all existing object detection models on real world datasets and is the first real-time model to achieve 60+ Average Precision on COCO talked w/ NVIDIA about it at GTC:

Joseph Nelson

18,231 次观看 • 1 年前

ZAI is on fire. Z.ai Hot on the heels of GLM-4.6V, the team has dropped two more models on Hugging Face: - AutoGLM-Phone-9B: a 9-billion-parameter “foundation model for smartphones” that reads the screen and acts on your behalf. Small enough for on-device inference, big enough to feel like future. - GLM-ASR-Nano-2512: a pocket-sized (2B) speech-recognition powerhouse that beats Whisper v3 on English, Mandarin, and Cantonese - It even picks up your whisper. Grab them here: - AutoGLM-Phone-9B → - GLM-ASR-Nano-2512 →

ZAI is on fire. Z.ai Hot on the heels of GLM-4.6V, the team has dropped two more models on Hugging Face: - AutoGLM-Phone-9B: a 9-billion-parameter “foundation model for smartphones” that reads the screen and acts on your behalf. Small enough for on-device inference, big enough to feel like future. - GLM-ASR-Nano-2512: a pocket-sized (2B) speech-recognition powerhouse that beats Whisper v3 on English, Mandarin, and Cantonese - It even picks up your whisper. Grab them here: - AutoGLM-Phone-9B → - GLM-ASR-Nano-2512 →

Tiezhen WANG

45,514 次观看 • 9 个月前

RF-DETR, the state-of-the-art model series for real-time object detection, can now run 100% locally in your browser on WebGPU with 🤗 Transformers.js v4! The models are Apache-2.0 licensed, making them a perfect fit for both personal and commercial applications. Try the demo 👇

RF-DETR, the state-of-the-art model series for real-time object detection, can now run 100% locally in your browser on WebGPU with 🤗 Transformers.js v4! The models are Apache-2.0 licensed, making them a perfect fit for both personal and commercial applications. Try the demo 👇

Xenova

77,179 次观看 • 7 个月前

Introducing Scribe — the most accurate Speech to Text model. It has the highest accuracy on benchmarks, outperforming previous state-of-the-art models such as Gemini 2.0 and OpenAI Whisper v3. It’s now the leading model for English, Spanish, Italian, and many more. With support for 99 languages, speaker diarization, character-level timestamps, and non-speech events such as laughing.

Introducing Scribe — the most accurate Speech to Text model. It has the highest accuracy on benchmarks, outperforming previous state-of-the-art models such as Gemini 2.0 and OpenAI Whisper v3. It’s now the leading model for English, Spanish, Italian, and many more. With support for 99 languages, speaker diarization, character-level timestamps, and non-speech events such as laughing.

ElevenLabs

464,967 次观看 • 1 年前

Introducing Arena Mode in Windsurf: One prompt. Two models. Your vote. Benchmarks don't reflect real-world coding quality. The best model for you depends on your codebase and stack. So we made real-world coding the benchmark. Free for the next week. May the best model win.

Introducing Arena Mode in Windsurf: One prompt. Two models. Your vote. Benchmarks don't reflect real-world coding quality. The best model for you depends on your codebase and stack. So we made real-world coding the benchmark. Free for the next week. May the best model win.

Devin Desktop

1,070,695 次观看 • 7 个月前

EmbeddingGemma is our new best-in-class open embedding model designed for on-device AI. 📱 At just 308M parameters, it delivers state-of-the-art performance while being small and efficient enough to run anywhere - even without an internet connection.

EmbeddingGemma is our new best-in-class open embedding model designed for on-device AI. 📱 At just 308M parameters, it delivers state-of-the-art performance while being small and efficient enough to run anywhere - even without an internet connection.

Google DeepMind

584,819 次观看 • 1 年前

InternLM-XComposer-2.5 A Versatile Large Vision Language Model Supporting Long-Contextual Input and Output We present InternLM-XComposer-2.5 (IXC-2.5), a versatile large-vision language model that supports long-contextual input and output. IXC-2.5 excels in various text-image comprehension and composition applications, achieving GPT-4V level capabilities with merely 7B LLM backend. Trained with 24K interleaved image-text contexts, it can seamlessly extend to 96K long contexts via RoPE extrapolation. This long-context capability allows IXC-2.5 to excel in tasks requiring extensive input and output contexts. Compared to its previous 2.0 version, InternLM-XComposer-2.5 features three major upgrades in vision-language comprehension: (1) Ultra-High Resolution Understanding, (2) Fine-Grained Video Understanding, and (3) Multi-Turn Multi-Image Dialogue. In addition to comprehension, IXC-2.5 extends to two compelling applications using extra LoRA parameters for text-image composition: (1) Crafting Webpages and (2) Composing High-Quality Text-Image Articles. IXC-2.5 has been evaluated on 28 benchmarks, outperforming existing open-source state-of-the-art models on 16 benchmarks. It also surpasses or competes closely with GPT-4V and Gemini Pro on 16 key tasks.

InternLM-XComposer-2.5 A Versatile Large Vision Language Model Supporting Long-Contextual Input and Output We present InternLM-XComposer-2.5 (IXC-2.5), a versatile large-vision language model that supports long-contextual input and output. IXC-2.5 excels in various text-image comprehension and composition applications, achieving GPT-4V level capabilities with merely 7B LLM backend. Trained with 24K interleaved image-text contexts, it can seamlessly extend to 96K long contexts via RoPE extrapolation. This long-context capability allows IXC-2.5 to excel in tasks requiring extensive input and output contexts. Compared to its previous 2.0 version, InternLM-XComposer-2.5 features three major upgrades in vision-language comprehension: (1) Ultra-High Resolution Understanding, (2) Fine-Grained Video Understanding, and (3) Multi-Turn Multi-Image Dialogue. In addition to comprehension, IXC-2.5 extends to two compelling applications using extra LoRA parameters for text-image composition: (1) Crafting Webpages and (2) Composing High-Quality Text-Image Articles. IXC-2.5 has been evaluated on 28 benchmarks, outperforming existing open-source state-of-the-art models on 16 benchmarks. It also surpasses or competes closely with GPT-4V and Gemini Pro on 16 key tasks.

AK

66,476 次观看 • 2 年前

Google presents AudioPaLM: A Large Language Model That Can Speak and Listen paper page: introduce AudioPaLM, a large language model for speech understanding and generation. AudioPaLM fuses text-based and speech-based language models, PaLM-2 [Anil et al., 2023] and AudioLM [Borsos et al., 2022], into a unified multimodal architecture that can process and generate text and speech with applications including speech recognition and speech-to-speech translation. AudioPaLM inherits the capability to preserve paralinguistic information such as speaker identity and intonation from AudioLM and the linguistic knowledge present only in text large language models such as PaLM-2. We demonstrate that initializing AudioPaLM with the weights of a text-only large language model improves speech processing, successfully leveraging the larger quantity of text training data used in pretraining to assist with the speech tasks. The resulting model significantly outperforms existing systems for speech translation tasks and has the ability to perform zero-shot speech-to-text translation for many languages for which input/target language combinations were not seen in training. AudioPaLM also demonstrates features of audio language models, such as transferring a voice across languages based on a short spoken prompt.

Google presents AudioPaLM: A Large Language Model That Can Speak and Listen paper page: introduce AudioPaLM, a large language model for speech understanding and generation. AudioPaLM fuses text-based and speech-based language models, PaLM-2 [Anil et al., 2023] and AudioLM [Borsos et al., 2022], into a unified multimodal architecture that can process and generate text and speech with applications including speech recognition and speech-to-speech translation. AudioPaLM inherits the capability to preserve paralinguistic information such as speaker identity and intonation from AudioLM and the linguistic knowledge present only in text large language models such as PaLM-2. We demonstrate that initializing AudioPaLM with the weights of a text-only large language model improves speech processing, successfully leveraging the larger quantity of text training data used in pretraining to assist with the speech tasks. The resulting model significantly outperforms existing systems for speech translation tasks and has the ability to perform zero-shot speech-to-text translation for many languages for which input/target language combinations were not seen in training. AudioPaLM also demonstrates features of audio language models, such as transferring a voice across languages based on a short spoken prompt.

AK

290,517 次观看 • 3 年前

New LLMs that control UIs! ByteDance Research releases UI-TARS, fine-tuned GUI agent that integrates reasoning, and action capabilities into a single vision-language model. Think of computer use but open. 👀 TL;DR; 3️⃣ Available in 3 sizes: 2B, 7B, and 72B parameters 🧠 Trained Qwen2-VL models with SFT & DPO 🥇 72B version achieves 82.8% on VisualWebBench (beating GPT-4 and Claude) 🏆 Achieves state-of-the-art results on 10+ GUI agent benchmarks 💡 Reasons before taking an action 🧑🏻‍💻 Can Click, Long Press, type, scroll, open app, navigate back/home, wait 🤗 Released under Apache 2.0 on Hugging Face

New LLMs that control UIs! ByteDance Research releases UI-TARS, fine-tuned GUI agent that integrates reasoning, and action capabilities into a single vision-language model. Think of computer use but open. 👀 TL;DR; 3️⃣ Available in 3 sizes: 2B, 7B, and 72B parameters 🧠 Trained Qwen2-VL models with SFT & DPO 🥇 72B version achieves 82.8% on VisualWebBench (beating GPT-4 and Claude) 🏆 Achieves state-of-the-art results on 10+ GUI agent benchmarks 💡 Reasons before taking an action 🧑🏻‍💻 Can Click, Long Press, type, scroll, open app, navigate back/home, wait 🤗 Released under Apache 2.0 on Hugging Face

Philipp Schmid

48,170 次观看 • 1 年前

people burn $1,900 a month on cloud, this $2,999 NVIDIA box runs 5 AI models at once and costs $10 to power 128GB of unified memory, stack two together and you double it, enough to run models a $2,000 graphics card cannot even open in the demo it ran Qwen3 Vision, Nemotron, a voice model, text to speech and speech to text all at once, and only used half the box here is the math that makes people switch: > cloud GPUs were costing $1,900 every single month, billed forever > the box is $2,999 one time, then about $10 of electricity a month > it paid for itself in two months, $22,000 stayed in the business in year one > zero data ever leaves your desk, no rental, no rate limits most people are still renting what this owns outright the full breakdown of the box and the exact math is in the article below

people burn $1,900 a month on cloud, this $2,999 NVIDIA box runs 5 AI models at once and costs $10 to power 128GB of unified memory, stack two together and you double it, enough to run models a $2,000 graphics card cannot even open in the demo it ran Qwen3 Vision, Nemotron, a voice model, text to speech and speech to text all at once, and only used half the box here is the math that makes people switch: > cloud GPUs were costing $1,900 every single month, billed forever > the box is $2,999 one time, then about $10 of electricity a month > it paid for itself in two months, $22,000 stayed in the business in year one > zero data ever leaves your desk, no rental, no rate limits most people are still renting what this owns outright the full breakdown of the box and the exact math is in the article below

John Doe

48,135 次观看 • 3 个月前

QVAC SDK 0.13.0 is live, and this version brings a lot of exciting updates! Local AI now plugs into your coding agent, ships as a desktop app in one command, and runs even more models. Highlights: NEW INTEGRATIONS - OpenCode and coding agents: the new @qvac/ai-sdk-provider makes QVAC a local provider. Less setup, same-model requests queue cleanly, and managed mode starts and supervises qvac serve for you. - Broader OpenAI-compatible API, validated across supported flows so covered capabilities stay consistent and testable. - Turn your QVAC project into a real desktop app for Mac, Windows, or Linux with a single command. The new Electron plugin handles the packaging and keeps the app small by including only what it needs. NEW MODELS - New pi0.5 model support - run a vision-language "robot brain" on a single ordinary graphics card, at full accuracy. - Image-to-video, fully local, via the Wan2.1 model in the Diffusion addon. - New BCI add-on: brain-computer interface transcription, fully local. Decode recorded neural signals into text on-device via the Whisper.cpp-based BCI model. IMPROVEMENTS - Whisper GPU transcription on Android, auto-picking the best backend (OpenCL on Adreno 700+, Vulkan elsewhere), unified on one ggml engine. - Parakeet steadier on mobile, with real end-of-utterance detection for streaming. - Supertonic TTS now runs full GPU across Metal, Vulkan, and OpenCL, with native streaming.

QVAC SDK 0.13.0 is live, and this version brings a lot of exciting updates! Local AI now plugs into your coding agent, ships as a desktop app in one command, and runs even more models. Highlights: NEW INTEGRATIONS - OpenCode and coding agents: the new @qvac/ai-sdk-provider makes QVAC a local provider. Less setup, same-model requests queue cleanly, and managed mode starts and supervises qvac serve for you. - Broader OpenAI-compatible API, validated across supported flows so covered capabilities stay consistent and testable. - Turn your QVAC project into a real desktop app for Mac, Windows, or Linux with a single command. The new Electron plugin handles the packaging and keeps the app small by including only what it needs. NEW MODELS - New pi0.5 model support - run a vision-language "robot brain" on a single ordinary graphics card, at full accuracy. - Image-to-video, fully local, via the Wan2.1 model in the Diffusion addon. - New BCI add-on: brain-computer interface transcription, fully local. Decode recorded neural signals into text on-device via the Whisper.cpp-based BCI model. IMPROVEMENTS - Whisper GPU transcription on Android, auto-picking the best backend (OpenCL on Adreno 700+, Vulkan elsewhere), unified on one ggml engine. - Parakeet steadier on mobile, with real end-of-utterance detection for streaming. - Supertonic TTS now runs full GPU across Metal, Vulkan, and OpenCL, with native streaming.

QVAC

20,932,195 次观看 • 3 个月前

📸 Introducing Replicate Image Editing Arena. We heard you needed an easy way to compare all of the latest image editing models. Evaluate models like FLUX.2, Nano Banana, Seedream, Qwen Edit, and Reve for speed, cost, and performance...all in one interface.

📸 Introducing Replicate Image Editing Arena. We heard you needed an easy way to compare all of the latest image editing models. Evaluate models like FLUX.2, Nano Banana, Seedream, Qwen Edit, and Reve for speed, cost, and performance...all in one interface.

Replicate

45,689 次观看 • 9 个月前