@rikas_ilamdeen Hi, you are the first one, NanoKVM is right, do you want NanoKVM or NanoKVM-Pro?
English
Sipeed
5.2K posts
Sipeed
@SipeedIO
AIoT opensource hardware platform
中华人民共和国 Katılım Eylül 2018
236 Takip Edilen24.1K Takipçiler
#Picoclaw Just hit 25K star~
We’re integrating PicoClaw into 3 offical hardware devices! 💻
Drop your guesses in the comments below—the first person to guess a device correctly will win it for FREE! 🎁👇 #openclaw
github.com/sipeed/picoclaw
English
#SLogic32U3 8CH capture @ 800 MHz — PASS ✅
Streaming bandwidth **800 MB/s** !
A New Era for Logic Analyzers Begins. 🚀
English
Tooling for the 4K AI Camera #MaixCAM2 enclosure is underway!
We expect to receive the samples next week, and we'll start shipping the week after next! 🚀
kickstarter.com/projects/zepan…
English
#NanoKVM & NanoKVM-Pro (WiFi Model) Security Update!
🐛 Fixed: Vulnerability during WiFi AP setup stage (local attacker in wifi range).
🙌 Thanks: Massive shoutout to the security experts who reported it. We’ll keep hardening NanoKVM! 🛡️
github.com/sipeed/NanoKVM…
English
这款手机大小的设备能运行120B参数的AI模型，无需0,000的Mac Studio。它支持本地化部署，确保100%隐私，可全天候驱动开源代理如OpenClaw，甚至替代聊天机器人。无需增加成本，即可进入AI模型的深水区。关注这个新趋势，看看未来谁会被淘汰。
中文
@SipeedIO guys are you gonna make another bunch of devkit kvms on nano board? I like it a lot. But I see r.n only pure edition
English
@no0ourrrr @TeksEdge @tiiny Normal AMD 8845HS miniPC(260$) + 96GB DDR5 (240$, 1 years ago, currently maybe 960$...).
English
🦾 Are you ready to carry your LLMs with you? This is the personal device of the future and potable personal inferencing is here from @tiiny! 💥 Models being updated.
🚀 Pocket AI beast specs:
🧠 CPU: ARMv9.2 12-core
⚡ AI Power: Custom SoC + dNPU, ~190 TOPS
💾 Memory/Storage: 80GB LPDDR5X RAM + 1TB SSD
🤖 Model Capacity: Runs up to 120B-parameter LLMs fully on-device
🔋 Efficiency: 30W TDP, ~65W typical system power
📏 Size/Weight: 14.2 × 8 × 2.53 cm, ~300g
🌐 Ecosystem: One-click deployment for dozens of open-source LLMs + agent frameworks
🔒 Connectivity: Fully offline — no internet or cloud needed
Supports:
🦙 Llama
🔮 Qwen
🌊 DeepSeek
🧩 Mistral
🪶 Phi
🌐 GLM
💎 Gemma
🧠 GPT-OSS
⚡ MiniCPM
🐉 Yi
🏛️ InternLM
🔥 Hunyuan
☁️ Skywork
📺 Bijan did a good video about it. Link in alt
English
Ok, but that means 48~64GB normal CPU solution is able to run this low-bit 120B model too...
In fact, we were planed to make a device that able to run dense 30B at the speed tiiny run A3B model. but the DDR price is too high now... but tiiny's ks result shocked us, we will consider it again!
English
@SipeedIO Yes they are. But they constrain the accuracy gap between quantized and original floating-point models to within ±0.5 percentage points.
English
You can run local AI models up to 120B without a $10,000 Mac Studio
This phone-sized device is your own server for open-source models.
100% local and private.
And you can use it to:
- Power an agent like OpenClaw 24/7
- Completely replace a chatbot
- Literally anything that requires an API
It’s called Tiiny, and, once again, you can run the latest open-source models on it.
Link below
English
Just curious about another thing: its 80GB of memory is actually split into 48GB dNPU and 32GB CPU. Typically, it's really hard to pool these two together effectively when running a single model. Since a 120B model at INT4 requires at least 60GB of memory, are they using an even lower-bit quantization?
English
@SipeedIO Yep can't agree more on the memory / bandwidth part. Honestly the benefit of are the massive amount of memory, the NPU usage so you it doesn't consume a lot of power and the form factor. But the MoE disclosure could be important, true.
English
Thank you for your tutorials!
Just want to clarify, their marketing is honestly a bit deceptive. They don't disclose the active parameters for any of the 3 models, tricking users into thinking they're dense. It only runs a 120B model because of the massive RAM, not because of some insane processing power.
BTW, I have a $500 8845 mini PC packed with 96GB LPDDR5X (snagged it 1 year ago during the memory price dip), so I can run 120B models too. For local LLM setups today, the real cost bottleneck is always memory capacity and bandwidth, not raw compute.
English
@SipeedIO Big fan of PicoClaw btw. Wrote an article to run it on Android. GPT OSS 120B has 5.1B active parameters not 3B + you still have to fit the full 120B in the RAM which is just not possible on your $500 mini PC... or my $1200 laptop with 32GB of RAM.
English
#ローカルLLM
なかなか興味深い🤔
もちろん #EVOX2 より遅いですが、省電力が魅力
David Hendrickson@TeksEdge
🦾 Are you ready to carry your LLMs with you? This is the personal device of the future and potable personal inferencing is here from @tiiny! 💥 Models being updated. 🚀 Pocket AI beast specs: 🧠 CPU: ARMv9.2 12-core ⚡ AI Power: Custom SoC + dNPU, ~190 TOPS 💾 Memory/Storage: 80GB LPDDR5X RAM + 1TB SSD 🤖 Model Capacity: Runs up to 120B-parameter LLMs fully on-device 🔋 Efficiency: 30W TDP, ~65W typical system power 📏 Size/Weight: 14.2 × 8 × 2.53 cm, ~300g 🌐 Ecosystem: One-click deployment for dozens of open-source LLMs + agent frameworks 🔒 Connectivity: Fully offline — no internet or cloud needed Supports: 🦙 Llama 🔮 Qwen 🌊 DeepSeek 🧩 Mistral 🪶 Phi 🌐 GLM 💎 Gemma 🧠 GPT-OSS ⚡ MiniCPM 🐉 Yi 🏛️ InternLM 🔥 Hunyuan ☁️ Skywork 📺 Bijan did a good video about it. Link in alt
日本語
@davideciffa @TeksEdge @tiiny Just tested, my 500$ 8845HS miniPC run faster than this 1500$ device, 32tps > 28tps, peoples are fooled...
They use the MOE model for promotion, 120B is only 3B actived, don't you feel strange that 120B,30B,20B model have almost same output eval speed?
English
@DegenApeDev @TeksEdge @tiiny Just tested, my 500$ 8845HS miniPC run faster than this 1500$ device, 32tps > 28tps, peoples are fooled...
They use the MOE model for promotion, 120B is only 3B actived, don't you feel strange that 120B,30B,20B model have almost same output eval speed?
English
@EtherCoins @TeksEdge Just tested, my 500$ 8845HS miniPC run faster than this 1500$ device, 32tps > 28tps, peoples are fooled..
English
I’m ready to buy this mobile GPU when it comes available. Qwen3.5 and Gemma4 along w/Stable Diffusion for Images on the road or at home. Private and subscription or API cost. It’s got 🦞 Clawdbot like functionality built in.
David Hendrickson@TeksEdge
🦾 Are you ready to carry your LLMs with you? This is the personal device of the future and potable personal inferencing is here from @tiiny! 💥 Models being updated. 🚀 Pocket AI beast specs: 🧠 CPU: ARMv9.2 12-core ⚡ AI Power: Custom SoC + dNPU, ~190 TOPS 💾 Memory/Storage: 80GB LPDDR5X RAM + 1TB SSD 🤖 Model Capacity: Runs up to 120B-parameter LLMs fully on-device 🔋 Efficiency: 30W TDP, ~65W typical system power 📏 Size/Weight: 14.2 × 8 × 2.53 cm, ~300g 🌐 Ecosystem: One-click deployment for dozens of open-source LLMs + agent frameworks 🔒 Connectivity: Fully offline — no internet or cloud needed Supports: 🦙 Llama 🔮 Qwen 🌊 DeepSeek 🧩 Mistral 🪶 Phi 🌐 GLM 💎 Gemma 🧠 GPT-OSS ⚡ MiniCPM 🐉 Yi 🏛️ InternLM 🔥 Hunyuan ☁️ Skywork 📺 Bijan did a good video about it. Link in alt
English
@itsPaulAi If that thing runs 120B locally, cloud bills just became optional overnight.
English