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。

Hunyuan-Large by Tencent is a 389B param MOE (52B active). It's the largest open-weights MOE. In some benchmarks it exceeds Llama 3.1 405B. With MLX's new 3-bit quant it just barely fits on a single 192GB M2 Ultra! And runs at a very decent >15 toks/sec:

Awni Hannun

34,562 subscribers

64,939 次观看 • 1 年前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

11 条评论

Awni Hannun 的头像

Awni Hannun1 年前

- Running it requires setting some sysctls. See the documentation here: - 3-bit model is in the HF MLX Community: - Sample command: `mlx_lm.generate --model mlx-community/Hunyuan-A52B-Instruct-3bit --prompt "Write a story about Einstein in English." -m 128 --system-prompt "You are a helpful AI assistant" --eos-token "<|eos|>"`

Awni Hannun 的头像

Awni Hannun1 年前

Thanks to @cognitivecompai for making the safetensors versions!

Xeophon 的头像

Xeophon1 年前

Its not even close to being the largest, Switch Transformers (2022) is 1.6T

Sam Hogan 🇺🇸 的头像

Sam Hogan 🇺🇸1 年前

@alexocheema video incoming

Ivan Fioravanti ᯅ 的头像

Ivan Fioravanti ᯅ1 年前

Oh wow! 😮

AM 的头像

AM1 年前

🤯

ankur 的头像

ankur1 年前

is there any publicly avail inference to try this out?

Reza Sayar 的头像

Reza Sayar1 年前

hmmm.. 3-bit quant you say? 👀how's the performance drop vs 4 or 8 bit?

BlurstTewiter 🌊🌊 的头像

BlurstTewiter 🌊🌊1 年前

Tencent = Hacking your machine DO NOT INSTALL

attunewise.ai 的头像

attunewise.ai1 年前

Can you run nemotron 340b with mlx on that type of machine?

Awni Hannun 的头像

Awni Hannun1 年前

It will fit in RAM in 3-bit.. but it will be a lot slower than the MOE. Something like 4-5x slower (so ~3 toks/sec).

相关视频

The new Deep Seek V3 0324 in 4-bit runs at > 20 toks/sec on a 512GB M3 Ultra with mlx-lm!

The new Deep Seek V3 0324 in 4-bit runs at > 20 toks/sec on a 512GB M3 Ultra with mlx-lm!

Awni Hannun

168,896 次观看 • 1 年前

DeepSeek R1 (the 680B MOE) is ~20% faster in the latest mlx / mlx-lm. 4-bit model on 3 M2 Ultras generates 4k tokens at a respectable 15 toks/sec. Plus some QoL improvements: - Only downloads the local shard (much faster startup) - Distributed launcher ships with MLX

DeepSeek R1 (the 680B MOE) is ~20% faster in the latest mlx / mlx-lm. 4-bit model on 3 M2 Ultras generates 4k tokens at a respectable 15 toks/sec. Plus some QoL improvements: - Only downloads the local shard (much faster startup) - Distributed launcher ships with MLX

Awni Hannun

86,921 次观看 • 1 年前

Tencent's Hy3 lands today with real coding and agent ability and a price that undercuts the models you'd put it against. It's a 295B MoE with only 21B active parameters, a big part of why it stays cheap to serve, plus a 256K context window and an Apache 2.0 license that even outperforms GLM-5.1 although GLM is 2.5x larger. On OpenRouter it runs at $0.14 per million input tokens and $0.58 per million output. For a model Tencent benchmarks against much larger flagships, that pricing is insane. Price-performance-ratio is China's real moat. I already wrote about Tencent Hy3 earlier and the significant leap in quality it represents. Therefore, I'm pleased to recommend this model from Tencent Hy. Open Router link down below: #Hy3 #Hunyuan #AICoding

Tencent's Hy3 lands today with real coding and agent ability and a price that undercuts the models you'd put it against. It's a 295B MoE with only 21B active parameters, a big part of why it stays cheap to serve, plus a 256K context window and an Apache 2.0 license that even outperforms GLM-5.1 although GLM is 2.5x larger. On OpenRouter it runs at $0.14 per million input tokens and $0.58 per million output. For a model Tencent benchmarks against much larger flagships, that pricing is insane. Price-performance-ratio is China's real moat. I already wrote about Tencent Hy3 earlier and the significant leap in quality it represents. Therefore, I'm pleased to recommend this model from Tencent Hy. Open Router link down below: #Hy3 #Hunyuan #AICoding

Chubby♨️

22,879 次观看 • 1 个月前

Imagine typing a sentence… and stepping into a world. At #WAIC2025, Tencent revealed the Hunyuan 3D World Model: ⮕ Create full 360° 3D scenes from just text ⮕ Explore in real time ⮕ Edit objects and backgrounds separately ⮕ Export to obj/.glb for game engines Pair that with: ⮕ Hunyuan-A13B — a lean, mean 13B-parameter MoE ⮕ Two open benchmark datasets for devs Hunyuan is lowering the barriers to a fully generative future. #Hunyuan3D #TencentAI #WAIC2025 #AIxCreation

Imagine typing a sentence… and stepping into a world. At #WAIC2025, Tencent revealed the Hunyuan 3D World Model: ⮕ Create full 360° 3D scenes from just text ⮕ Explore in real time ⮕ Edit objects and backgrounds separately ⮕ Export to obj/.glb for game engines Pair that with: ⮕ Hunyuan-A13B — a lean, mean 13B-parameter MoE ⮕ Two open benchmark datasets for devs Hunyuan is lowering the barriers to a fully generative future. #Hunyuan3D #TencentAI #WAIC2025 #AIxCreation

SANI BULA

34,630 次观看 • 1 年前

You can now try Llama 3.1 405B for free (link below)! This is the largest open-source model out there, and for the first time, an open model is competitive with closed models. This time around, Meta did something new: Llama 3.1 has a license that allows developers to use it to enhance other models. For the first time, you can distill Llama 3.1 405B's capabilities into a smaller, more practical model for your use case. First, here is the link where you can play with Llama 3.1 for free: The model is hosted in Tune Studio, an end-to-end platform for developing applications using Large Language Models. They are sponsoring this post. Take a look at the attached video. It will show you how you can fine-tune a simple model using Llama 3.1 without leaving the platform: 1. You can create an empty dataset 2. Use the playground to generate and record interactions with Llama 3.1 3. Modify the dataset directly using the playground 4. Export the data and fine-tune a smaller model Fast and easy! As long as you have a web browser, you can start experimenting with fine-tuning and Llama 3.1. That's all it takes!

You can now try Llama 3.1 405B for free (link below)! This is the largest open-source model out there, and for the first time, an open model is competitive with closed models. This time around, Meta did something new: Llama 3.1 has a license that allows developers to use it to enhance other models. For the first time, you can distill Llama 3.1 405B's capabilities into a smaller, more practical model for your use case. First, here is the link where you can play with Llama 3.1 for free: The model is hosted in Tune Studio, an end-to-end platform for developing applications using Large Language Models. They are sponsoring this post. Take a look at the attached video. It will show you how you can fine-tune a simple model using Llama 3.1 without leaving the platform: 1. You can create an empty dataset 2. Use the playground to generate and record interactions with Llama 3.1 3. Modify the dataset directly using the playground 4. Export the data and fine-tune a smaller model Fast and easy! As long as you have a web browser, you can start experimenting with fine-tuning and Llama 3.1. That's all it takes!

Santiago

55,609 次观看 • 2 年前

🤯Laguna S-2.1 UD-IQ3_S on a single 3090 ~30-35t/s decode ~280-290t/s prefill ~23GB VRAM @ 64k context, f16 KV llama.cpp: -ngl 999 + --n-cpu-moe 32 + fa on 118B MoE, ~8B active, did we get a decent mid-range model that doesn't need multi gpu and can run with good enough speeds for long-context coding tasks? Might actually stick with this one. Been waiting for a mid-range model that fits on a single 3090, this is it. Usual three.js tests underway. Update soon.

🤯Laguna S-2.1 UD-IQ3_S on a single 3090 ~30-35t/s decode ~280-290t/s prefill ~23GB VRAM @ 64k context, f16 KV llama.cpp: -ngl 999 + --n-cpu-moe 32 + fa on 118B MoE, ~8B active, did we get a decent mid-range model that doesn't need multi gpu and can run with good enough speeds for long-context coding tasks? Might actually stick with this one. Been waiting for a mid-range model that fits on a single 3090, this is it. Usual three.js tests underway. Update soon.

AJ

20,031 次观看 • 17 天前

Most instructors missed that the golf swing is actually simple, and Moe Norman proved it every time he hit a ball. Moe understood the club is really just a 3-foot hammer—weight on the end of a stick. While most golfers set up with their arms disconnected, Moe lined the club up with his arms, creating one straight line. That’s the beauty of the Single Plane Swing. Take it back… bring it straight back down. No wasted motion. No extra moving parts. Just simple, natural, and repeatable. Moe figured out the most natural way for the human body to strike a golf ball. #MoeNorman #SinglePlaneSwing #GolfTips #BallStriking #ToddGraves

Most instructors missed that the golf swing is actually simple, and Moe Norman proved it every time he hit a ball. Moe understood the club is really just a 3-foot hammer—weight on the end of a stick. While most golfers set up with their arms disconnected, Moe lined the club up with his arms, creating one straight line. That’s the beauty of the Single Plane Swing. Take it back… bring it straight back down. No wasted motion. No extra moving parts. Just simple, natural, and repeatable. Moe figured out the most natural way for the human body to strike a golf ball. #MoeNorman #SinglePlaneSwing #GolfTips #BallStriking #ToddGraves

Todd Graves

86,777 次观看 • 3 个月前

running a 405-billion-parameter model used to mean a $30,000 GPU rack. one developer made it run on a single 8GB card the kind that's been sitting in gaming laptops for years. it's called AirLLM, and the trick is almost stupidly simple: instead of loading the whole model, it keeps one layer on the GPU at a time and streams the rest through. / Llama 3.1 405B on 8GB / DeepSeek-V3, all 671B of it, on about 12GB / no quantization, no distillation, no pruning the full-precision model / works with Llama, Qwen, DeepSeek, Mistral, Gemma, and more it's Apache 2.0 and already past 22,600 stars. → bookmark it.

running a 405-billion-parameter model used to mean a $30,000 GPU rack. one developer made it run on a single 8GB card the kind that's been sitting in gaming laptops for years. it's called AirLLM, and the trick is almost stupidly simple: instead of loading the whole model, it keeps one layer on the GPU at a time and streams the rest through. / Llama 3.1 405B on 8GB / DeepSeek-V3, all 671B of it, on about 12GB / no quantization, no distillation, no pruning the full-precision model / works with Llama, Qwen, DeepSeek, Mistral, Gemma, and more it's Apache 2.0 and already past 22,600 stars. → bookmark it.

dunik

148,663 次观看 • 25 天前

🚀 Introducing HunyuanVideo-Avatar, a model jointly developed by Tencent Hunyuan and Tencent Music, bringing photos to life. ✅ Upload a photo + audio — auto-detect scene context & emotion, then generate lifelike speech/singing with dynamic visuals. ✅ Supports multi-style, multi-species scenarios and excels in multi-character interactions. ✅ Built for short video creation, e-commerce, advertising, and more — already deployed in multiple apps across Tencent Music Entertainment Group. The single-character mode of HunyuanVideo-Avatar is now open-sourced and live on the Hunyuan website, supporting audio up to 14 seconds for video generation. The multi-character mode will be open-sourced soon. Try it on: Project Page: Github: Tech report: #Tencent #Hunyuan

Tencent Hy

218,408 次观看 • 1 年前

six months ago this wasn't happening on 8gb vram. running unsloth's Q4_K_XL quant of gemma 4 26b-a4b-it-qat, a sparse MoE model with only 4b active params on a single rtx 4060 laptop gpu, 8gb vram, 20+ tok/s decode. no cloud, no api, no offload hacks. just a gaming laptop on battery. what makes it fit: google's QAT (quantization aware training), plus MTP (multi token prediction) support in the latest llama.cpp builds. that combo is the single biggest unlock for local inference on low vram. rtx 3060, rtx 3070, gtx 1070, gtx 1080, rtx 4050, rtx 4060, rtx 5050, rtx 5060 — any 6-8gb consumer gpu, old or new — this model runs on it. world cup season, so i told it to build a soccer themed flappy bird clone. one shot, zero iteration, fully playable. six months ago an 8gb model could barely clone vanilla flappy bird. now it's shipping a themed game from a sparse MoE model running locally on a laptop battery. inference benchmarks: - decode throughput: 30 tok/s - context: 64k. this is the real unlock. 64k ctx is what makes a hermes agent loop viable locally on this model, not just single-turn chat. llama.cpp flags: -m gemma-4-26B-A4B-it-qat-UD-Q4_K_XL.gguf -c 64000 -cmoe --port 8080 game's deployed on my own site, built and shipped end to end with open source llm, zero closed source api dependency in the pipeline. link in the description. gguf weights on huggingface, link in the comments. pull it down, run it on whatever 8gb card is sitting in your rig. try the game and tell me your score and what you want in v2. local llms on consumer gpus stopped being a meme.

six months ago this wasn't happening on 8gb vram. running unsloth's Q4_K_XL quant of gemma 4 26b-a4b-it-qat, a sparse MoE model with only 4b active params on a single rtx 4060 laptop gpu, 8gb vram, 20+ tok/s decode. no cloud, no api, no offload hacks. just a gaming laptop on battery. what makes it fit: google's QAT (quantization aware training), plus MTP (multi token prediction) support in the latest llama.cpp builds. that combo is the single biggest unlock for local inference on low vram. rtx 3060, rtx 3070, gtx 1070, gtx 1080, rtx 4050, rtx 4060, rtx 5050, rtx 5060 — any 6-8gb consumer gpu, old or new — this model runs on it. world cup season, so i told it to build a soccer themed flappy bird clone. one shot, zero iteration, fully playable. six months ago an 8gb model could barely clone vanilla flappy bird. now it's shipping a themed game from a sparse MoE model running locally on a laptop battery. inference benchmarks: - decode throughput: 30 tok/s - context: 64k. this is the real unlock. 64k ctx is what makes a hermes agent loop viable locally on this model, not just single-turn chat. llama.cpp flags: -m gemma-4-26B-A4B-it-qat-UD-Q4_K_XL.gguf -c 64000 -cmoe --port 8080 game's deployed on my own site, built and shipped end to end with open source llm, zero closed source api dependency in the pipeline. link in the description. gguf weights on huggingface, link in the comments. pull it down, run it on whatever 8gb card is sitting in your rig. try the game and tell me your score and what you want in v2. local llms on consumer gpus stopped being a meme.

Alok

60,866 次观看 • 1 个月前

It took us a bit, but Iron Lung is finally available on Android devices! The port was done by John Szymanski (Ducky) and I, so we can confirm that it plays well and runs well. With a pair of headphones, it's actually a decent way to play the game!

It took us a bit, but Iron Lung is finally available on Android devices! The port was done by John Szymanski (Ducky) and I, so we can confirm that it plays well and runs well. With a pair of headphones, it's actually a decent way to play the game!

David Szymanski

142,504 次观看 • 7 个月前

GLM 4.6 runs quite fast on an M3 Ultra with mlx-lm even at higher precision. Pretty remarkable that it benchmarks competitive to the just-released Sonnet 4.5. Hope those benchmarks hold-up in day-to-day use. Here's a run using 5.5 bpw quantized model, generating 5.3k tokens at 17+ tok/sec using 244 GB. What prompts should I test?

GLM 4.6 runs quite fast on an M3 Ultra with mlx-lm even at higher precision. Pretty remarkable that it benchmarks competitive to the just-released Sonnet 4.5. Hope those benchmarks hold-up in day-to-day use. Here's a run using 5.5 bpw quantized model, generating 5.3k tokens at 17+ tok/sec using 244 GB. What prompts should I test?

Awni Hannun

68,539 次观看 • 10 个月前

Exclusive: Meta just released Llama 3.1 405B — the first-ever open-sourced frontier AI model, beating top closed models like GPT-4o across several benchmarks. I sat down with Mark Zuckerberg, diving into why this marks a major moment in AI history. Timestamps: 00:00 Intro 00:38 Meta’s Llama 3.1 rundown 03:44 Real-world use cases for Llama 3.1 06:15 Educating developers on open-source AI tools 09:43 Societal implications of open-source AI 13:00 Balancing power and managing bad actors 14:40 Open source and global competition 16:59 Accelerating innovation and economic growth 20:04 Zuck on Apple and lessons from the past 24:22 Future of AI: Llama 3 and beyond 26:43 Prediction: Billions of personalized AI agents 31:32 Factors to changing anti-AI sentiment

Exclusive: Meta just released Llama 3.1 405B — the first-ever open-sourced frontier AI model, beating top closed models like GPT-4o across several benchmarks. I sat down with Mark Zuckerberg, diving into why this marks a major moment in AI history. Timestamps: 00:00 Intro 00:38 Meta’s Llama 3.1 rundown 03:44 Real-world use cases for Llama 3.1 06:15 Educating developers on open-source AI tools 09:43 Societal implications of open-source AI 13:00 Balancing power and managing bad actors 14:40 Open source and global competition 16:59 Accelerating innovation and economic growth 20:04 Zuck on Apple and lessons from the past 24:22 Future of AI: Llama 3 and beyond 26:43 Prediction: Billions of personalized AI agents 31:32 Factors to changing anti-AI sentiment

Rowan Cheung

2,652,458 次观看 • 2 年前

watch anon! 27 billion parameter model dancing at 42 tokens a second on a used rtx 3060ti, that costs 200 dollars. consuming 159 watts. temp 62 degrees. pushing almost 100% on the gpu. 6.8 of 8 gigs used with a 128k context window loaded. and every one of those spikes is the model reasoning and firing tool calls. the model is bonsai 27b at 1 bit. PrismML took qwen 3.6 27b, and crushed it to 3.9 gigabytes, about a single sign bit per weight. it drops onto an 8gb rtx 3060 ti, an ampere card from 2020, and it doesn't just load, it serves, it holds the full 128k window, it runs a hermes agent for half an hour straight, it hits its own failures and recovers, it never once drops the loop. the absurd part is the power meter. that same 27b, in its normal quant 4, wants a 24gb card and nearly 17 gigs of weights. this is it at 3.9, doing agentic work at 160 watts, on the tier people call budget. the numbers are sitting right there on nvtop, not in a slide, not a claim, the meter reading while it works. so i keep landing on one question. the model is open, apache 2.0. the card is old and cheap. the whole thing runs offline. why is almost nobody actually running this. is it the hardware, the setup, or did nobody tell you a 27b fits on the card already in your closet.

watch anon! 27 billion parameter model dancing at 42 tokens a second on a used rtx 3060ti, that costs 200 dollars. consuming 159 watts. temp 62 degrees. pushing almost 100% on the gpu. 6.8 of 8 gigs used with a 128k context window loaded. and every one of those spikes is the model reasoning and firing tool calls. the model is bonsai 27b at 1 bit. PrismML took qwen 3.6 27b, and crushed it to 3.9 gigabytes, about a single sign bit per weight. it drops onto an 8gb rtx 3060 ti, an ampere card from 2020, and it doesn't just load, it serves, it holds the full 128k window, it runs a hermes agent for half an hour straight, it hits its own failures and recovers, it never once drops the loop. the absurd part is the power meter. that same 27b, in its normal quant 4, wants a 24gb card and nearly 17 gigs of weights. this is it at 3.9, doing agentic work at 160 watts, on the tier people call budget. the numbers are sitting right there on nvtop, not in a slide, not a claim, the meter reading while it works. so i keep landing on one question. the model is open, apache 2.0. the card is old and cheap. the whole thing runs offline. why is almost nobody actually running this. is it the hardware, the setup, or did nobody tell you a 27b fits on the card already in your closet.

Sudo su

23,225 次观看 • 14 天前

This is the fastest I've seen Llama 3.3 running anywhere! Llama 3.3 70B running at 652 t/s is lightning fast. And if you want Llama 3.1, here are the speeds I was able to get: • Llama 3.1 8B: 1006 t/s • Llama 3.1 70B: 709 t/s • Llama 3.1 405B: 206 t/s (You can access all of these models for free! See the link below.) This speed is incredible, but the interesting part is what's happening behind the scenes: These models aren't running on a GPU! In this video, I'm using the SambaNova cloud to access these models. They built a custom chip (SN40L) optimized for AI workflows. A single SN40L chip can hold hundreds of models (trillions of parameters) in memory! The speed alone is a huge deal, but the big advantage is for agentic workflows running multiple specialized models. A GPU can only host a single model and switch (unload and load) to a different model if necessary. An SN40L, on the other hand, can host every model at once, making it much faster. Here is the video where you can see how fast these chips are:

This is the fastest I've seen Llama 3.3 running anywhere! Llama 3.3 70B running at 652 t/s is lightning fast. And if you want Llama 3.1, here are the speeds I was able to get: • Llama 3.1 8B: 1006 t/s • Llama 3.1 70B: 709 t/s • Llama 3.1 405B: 206 t/s (You can access all of these models for free! See the link below.) This speed is incredible, but the interesting part is what's happening behind the scenes: These models aren't running on a GPU! In this video, I'm using the SambaNova cloud to access these models. They built a custom chip (SN40L) optimized for AI workflows. A single SN40L chip can hold hundreds of models (trillions of parameters) in memory! The speed alone is a huge deal, but the big advantage is for agentic workflows running multiple specialized models. A GPU can only host a single model and switch (unload and load) to a different model if necessary. An SN40L, on the other hand, can host every model at once, making it much faster. Here is the video where you can see how fast these chips are:

Santiago

97,543 次观看 • 1 年前

MolmoAct2 runs zero-shot on an SO-ARM101, no training required 🦾 Ai2 open Action Reasoning Model is now in LeRobot with the full lifecycle covered: fine-tuning (full or LoRA), evaluation, and real-robot deployment. It pairs a Molmo2-ER vision-language backbone with a flow-matching action expert to turn images, language, and proprioceptive state into action chunks. The best part: ready-made checkpoints ship with calibration correction baked in, so you can point it at your SO-100/101 and just watch it work. Inference fits in ~12GB at bf16, and LoRA fine-tuning runs on a single 24GB GPU. Big thanks to the Ai2 team for building this in the open.

MolmoAct2 runs zero-shot on an SO-ARM101, no training required 🦾 Ai2 open Action Reasoning Model is now in LeRobot with the full lifecycle covered: fine-tuning (full or LoRA), evaluation, and real-robot deployment. It pairs a Molmo2-ER vision-language backbone with a flow-matching action expert to turn images, language, and proprioceptive state into action chunks. The best part: ready-made checkpoints ship with calibration correction baked in, so you can point it at your SO-100/101 and just watch it work. Inference fits in ~12GB at bf16, and LoRA fine-tuning runs on a single 24GB GPU. Big thanks to the Ai2 team for building this in the open.

LeRobot

22,200 次观看 • 1 个月前

JUST IN: Google releases Gemini 1.5, a powerful MoE model. It's a huge breakthrough. The model has the longest context window ever seen: 1 million tokens. It can process 1 hour of video, 11 hours of audio, 30,000 lines of code, or 700,000 words in a single prompt. When tested on text, code, image, audio and video evaluations, 1.5 Pro outperforms 1.0 Pro on 87% of the benchmarks used for developing LLMs. You can can sign up in AI Studio to try it out.

JUST IN: Google releases Gemini 1.5, a powerful MoE model. It's a huge breakthrough. The model has the longest context window ever seen: 1 million tokens. It can process 1 hour of video, 11 hours of audio, 30,000 lines of code, or 700,000 words in a single prompt. When tested on text, code, image, audio and video evaluations, 1.5 Pro outperforms 1.0 Pro on 87% of the benchmarks used for developing LLMs. You can can sign up in AI Studio to try it out.

Lior Alexander

83,432 次观看 • 2 年前

Tencent just dropped Hy3, and it's worth a look if you're building AI agents. I spent some time putting it through its paces today. A quick rundown of what makes this release notable: → 295B total parameters (Mixture-of-Experts), but only 21B active at inference — a real efficiency play → 256K context window for handling large codebases or long documents → Purpose-built improvements for coding and multi-step agent tasks → Fully open under Apache 2.0 — no restrictive commercial terms → A free two-week API window currently live on OpenRouter On the practical side: prompts involving layered instructions and coding tasks came back fast and coherent. Tencent says this release builds directly on feedback from 50+ internal product teams following an earlier preview, with reported drops in hallucination rate (12.5% → 5.4%) and commonsense errors (25.4% → 12.7%). Those are Tencent's own numbers, so treat them as a starting point rather than gospel until third parties weigh in. For context: Tencent's own comparisons put Hy3 behind GLM-5.2 specifically on coding benchmarks — GLM is a much bigger model (~744B total), so the trade-off makes sense. Hy3's pitch isn't "biggest," it's "efficient enough to actually deploy." Bottom line — if you're evaluating open-weight options for agent workloads, this is a solid one to add to the testing queue while the free window is open. Try it here: Tencent Hy #Hy3 #Hunyuan #TencentAI #AICoding

Tencent just dropped Hy3, and it's worth a look if you're building AI agents. I spent some time putting it through its paces today. A quick rundown of what makes this release notable: → 295B total parameters (Mixture-of-Experts), but only 21B active at inference — a real efficiency play → 256K context window for handling large codebases or long documents → Purpose-built improvements for coding and multi-step agent tasks → Fully open under Apache 2.0 — no restrictive commercial terms → A free two-week API window currently live on OpenRouter On the practical side: prompts involving layered instructions and coding tasks came back fast and coherent. Tencent says this release builds directly on feedback from 50+ internal product teams following an earlier preview, with reported drops in hallucination rate (12.5% → 5.4%) and commonsense errors (25.4% → 12.7%). Those are Tencent's own numbers, so treat them as a starting point rather than gospel until third parties weigh in. For context: Tencent's own comparisons put Hy3 behind GLM-5.2 specifically on coding benchmarks — GLM is a much bigger model (~744B total), so the trade-off makes sense. Hy3's pitch isn't "biggest," it's "efficient enough to actually deploy." Bottom line — if you're evaluating open-weight options for agent workloads, this is a solid one to add to the testing queue while the free window is open. Try it here: Tencent Hy #Hy3 #Hunyuan #TencentAI #AICoding

Felix

36,925 次观看 • 1 个月前

Introducing "Building with Llama 4." This short course is created with Meta AI at Meta, and taught by Amit Sangani, Director of Partner Engineering for Meta’s AI team. Meta’s new Llama 4 has added three new models and introduced the Mixture-of-Experts (MoE) architecture to its family of open-weight models, making them more efficient to serve. In this course, you’ll work with two of the three new models introduced in Llama 4. First is Maverick, a 400B parameter model, with 128 experts and 17B active parameters. Second is Scout, a 109B parameter model with 16 experts and 17B active parameters. Maverick and Scout support long context windows of up to a million tokens and 10M tokens, respectively. The latter is enough to support directly inputting even fairly large GitHub repos for analysis! In hands-on lessons, you’ll build apps using Llama 4’s new multimodal capabilities including reasoning across multiple images and image grounding, in which you can identify elements in images. You’ll also use the official Llama API, work with Llama 4’s long-context abilities, and learn about Llama’s newest open-source tools: its prompt optimization tool that automatically improves system prompts and synthetic data kit that generates high-quality datasets for fine-tuning. If you need an open model, Llama is a great option, and the Llama 4 family is an important part of any GenAI developer's toolkit. Through this course, you’ll learn to call Llama 4 via API, use its optimization tools, and build features that span text, images, and large context. Please sign up here:

Introducing "Building with Llama 4." This short course is created with Meta AI at Meta, and taught by Amit Sangani, Director of Partner Engineering for Meta’s AI team. Meta’s new Llama 4 has added three new models and introduced the Mixture-of-Experts (MoE) architecture to its family of open-weight models, making them more efficient to serve. In this course, you’ll work with two of the three new models introduced in Llama 4. First is Maverick, a 400B parameter model, with 128 experts and 17B active parameters. Second is Scout, a 109B parameter model with 16 experts and 17B active parameters. Maverick and Scout support long context windows of up to a million tokens and 10M tokens, respectively. The latter is enough to support directly inputting even fairly large GitHub repos for analysis! In hands-on lessons, you’ll build apps using Llama 4’s new multimodal capabilities including reasoning across multiple images and image grounding, in which you can identify elements in images. You’ll also use the official Llama API, work with Llama 4’s long-context abilities, and learn about Llama’s newest open-source tools: its prompt optimization tool that automatically improves system prompts and synthetic data kit that generates high-quality datasets for fine-tuning. If you need an open model, Llama is a great option, and the Llama 4 family is an important part of any GenAI developer's toolkit. Through this course, you’ll learn to call Llama 4 via API, use its optimization tools, and build features that span text, images, and large context. Please sign up here:

Andrew Ng

67,846 次观看 • 1 年前

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.

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

213,329 次观看 • 2 个月前