The future of AI should not be defined only by who builds the biggest models. It should also be defined by who gives more people the freedom to use them.
That is why @169Pi_ai deserves attention for Alpie Seed 9B.
Alpie Seed is a nine-billion-parameter, open-weight vision-language model built on Qwen3.5-9B and post-trained by 169Pi for reasoning, coding and instruction following. It accepts text, images and video, supports thinking and non-thinking modes, and has a native context window of 262,144 tokens.
Those are meaningful technical foundations. But the part that stands out to me is what the team is trying to make possible beyond the specification sheet: AI that people can deploy closer to where they actually work.
169Pi reports offline testing across laptops, workstations and smartphones. The release also includes 2-bit, 3-bit and 4-bit GGUF versions, giving developers options for different hardware and memory constraints. Performance depends on the device, runtime and model version-but the ability to run locally is the point.
Cloud services can be valuable. They should not have to be the only option.
A developer should be able to experiment on their own machine. A team should have the option to process information within its own environment. An application should be able to offer useful AI even when connectivity is limited.
That is the opportunity local deployment creates: more choice about where computation happens, how systems are integrated and who controls the infrastructure.
The Apache 2.0 release matters here. Downloadable weights give developers something tangible to evaluate, adapt and build into their own applications, subject to the licence terms. The relationship can go beyond using a service: builders can work directly with the model.
I also appreciate that the published documentation identifies the model's Qwen foundation, explains the post-training work and discusses deployment limitations. That is the detail serious builders need when deciding what to test-not just a headline score.
For me, the case for paying attention to Alpie Seed is not that one release has solved every challenge in AI. It is that capability, local deployment and developer freedom are being brought together in a release people can examine for themselves.
The next step is the interesting one: test it against real tasks, explore its limits and see what can be built with it.
There is room to celebrate larger, more capable models. There should be just as much room to celebrate work that makes capable models available in more places, under more people's control.
Respect to the team at 169Pi for contributing to that direction.
Alpie Seed 9B is a release worth looking at closely-not only for the model itself, but for the freedom to build around it.
The future of AI should have more builders, not just more users.
Respect to @169Pi_ai for the work behind Alpie Seed 9B-and for giving developers something they can explore, evaluate and build with.
#AI #EdgeAI #169Pi
Let's Write | Chat App retweeted
Those who see the vision ahead know!
solana:aqQRs3UJmHs7ktoLo46HZAzvG9P3oCoZ41nENqFbory is a utility token, not a meme token, that is here to stay.
Long-term!🤝
#PIE #Web3
Risk to reward here is phenomenal!
@169Pi_ai have only gotten 10x better from the previous ATH of $4.16M. Really take the time to look what's happened since. $PIE is bottomed and if not $135K MC is the asymmetric bet!. With a 36x back to ATH. These guys aren't going anywhere and are a highly decorated team.
Let's Write | Chat App retweeted
A $PIE was being cooked indeed 🥧
While there is significant hype for AI models and @AnthropicAI , @169Pi_ai $PIE is creating the most deployable, efficient and most optimized local/offline model
They are early to this new market and have a big research team behind it
Very clean
Bullish divergence forming that previously brought $PIE up to 4M
Something big is brewing for $PIE imo
@169Pi_ai
@PIEby169Pi
Let's Write | Chat App retweeted
Risk to reward here is phenomenal!
@169Pi_ai have only gotten 10x better from the previous ATH of $4.16M. Really take the time to look what's happened since. $PIE is bottomed and if not $135K MC is the asymmetric bet!. With a 36x back to ATH. These guys aren't going anywhere and are a highly decorated team.
A 9B model that can see, reason, use tools and run locally.
AI is becoming multimodal. But making multimodal intelligence practical beyond the cloud is a different engineering challenge.
That's what we're working on with Alpie Seed at 169Pi.
Alpie Seed brings together a 262,144-token context window, image and video understanding, and tool calling in a single model.
And we're working to make it deployable across the hardware people already have.
With Q4, Q3 and Q2 quantized versions, we have tested local inference across smartphones, Apple Silicon, NVIDIA GPUs and even CPU-only Windows systems.
The results go beyond simply getting a model to load. In our evaluations, Alpie Seed scored:
90.30 on OCRBench
68.70 on MMMU-Pro
76.76 on MathVision
80.39 on RealWorldQA
On an iPhone 17 Pro, our local inference test recorded 231 tokens/sec prompt processing and 9.59 tokens/sec generation, with 7 GB peak memory usage.
What does this make possible?
Developers building AI applications that can understand visual inputs and use tools. Users analysing documents, images and videos without relying on a cloud connection. Enterprises exploring AI deployments closer to their own data and infrastructure.
We're continuing to improve Alpie Seed, focusing on better accuracy, faster inference and greater efficiency across devices.
Long context. Multimodal intelligence. Tool use. Local execution.
One model, more places to put it to work.
Alpie Seed by 169Pi AI.
Let's Write | Chat App retweeted
For real tho!. There are 2 SOL utility tokens that have been building proprietary AI products since the beginning of the year, shipping huge achievements, hitting milestones and show up every day. Both are sub $1M in market cap.
$PIE - @169Pi_ai
Private Edge AI models. Extreme quantization & Compression of open models without losing reasoning. Extremely efficient cutting VRM significantly.
$BioLLM - @BioLLM_
BioLLM Studio is middleware for biological computing, connecting to living neurons and neural circuits. For thought to text interfaces, robotics, adaptave compute. Actually really worth diving IMO. Should be a billion dollar company!.
Let's stop scouting every new launch and putting money in the highly skilled larp devs hands that can dress a project quiet well these days and back some certified mother fucking builders.
Let's Write | Chat App retweeted
Very clean
Bullish divergence forming that previously brought $PIE up to 4M
Something big is brewing for $PIE imo
@169Pi_ai
@PIEby169Pi
Intelligence shouldn’t require an internet connection or a cloud bill on every prompt.
Watch Alpie Seed 9B read and reason over a full restaurant menu from a photo on an iPhone in Airplane Mode. Zero Wi-Fi. Zero cellular. Zero data leaving the device.
4-bit quantization is doing for AI what MP3 did for music: shrinking massive multimodal intelligence down to a footprint that fits in your pocket without sacrificing real-world utility.
Here is what’s running locally on that phone:
• Architecture: 9B multimodal reasoning model (text, image, video) with a 262K context window
• Footprint: ~5GB in 4-bit GGUF runs smoothly on iPhones, Android flagships, and consumer laptops
• Post-Training: Built on Qwen3.5-9B via a two-stage SFT + GRPO pipeline with DoRA adapters, trained with 4-bit NF4 quantization and BF16 compute for edge-ready efficiency
• Cost: $0 per token, forever
• License: 100% open weights (Apache 2.0)
You don't need a cluster to test it. Run it on your machine right now in one command: ollama run ollama.com/169pi/alpie-seed
We’re building sovereign, device-native AI for the next billion devices from phones in airplane mode to air-gapped enterprise workflows.
Drop your device specs + tokens/sec in the replies if you run it locally 👇
A robot doesn't need a massive model to navigate a room. Your phone doesn't need a data centre to understand a document.
Different devices. Different constraints. Different jobs.
So why should they all depend on the same kind of AI?
With Alpie Seed, we're exploring multimodal intelligence across phones, laptops, workstations, robotics and edge devices.
The goal isn't to put the biggest model everywhere.
It's to make capable AI work where it's actually needed.
One model. More ways to put intelligence to work.
Explore the model here:
@huggingface: huggingface.co/169Pi/Alpie_s…
GGUF: huggingface.co/169Pi/Alpie_S…
@ollama: ollama.com/169pi/alpie-seed
@kaggle: kaggle.com/models/alpie/alpi…
Some messages aren't about making plans or solving something.
They're just there to say:
I'm thinking of you.
How are you really doing?
I hope you're okay.
We're building Let's Write for the kinds of conversations that keep people close.
What kind of message means the most to you?






















