This Diwali, a little more light from India's AI research community onto the global AI landscape.
Introducing the latest benchmark results from Alpie Seed, our 9B multimodal reasoning model developed at @169Pi_ai Research.
We believe the next generation of AI will be shaped not only by models that scale bigger, but by the research and engineering that make smaller models increasingly capable.
We put Alpie Seed through a range of demanding evaluations across reasoning, scientific knowledge, mathematics, coding and software engineering.
Here are some of the results:
→ 82.54% on MMLU-Pro
→ 84.32% on GPQA Diamond
→ 72.99% on LiveCodeBench v6
→ 61.20% on SWE-Bench Verified
The comparison highlights some interesting results.
Alpie Seed records the highest GPQA Diamond score among the models shown. On MMLU-Pro, it performs competitively with larger models, while its LiveCodeBench result is within striking distance of substantially larger systems.
Its SWE-Bench Verified result of 61.20% is another important milestone, reflecting its ability to work through real-world software engineering tasks. This evaluation used mini-SWE-agent without retries.
Beyond benchmark accuracy, we also examined inference efficiency.
In a separate profiling evaluation, Alpie Seed recorded:
→ 36.66 ms time to first token
→ 41.37 tokens/s generation speed
→ 9,796.22 tokens/s prompt processing speed
These measurements come from a separate profiling setup and should not be confused with the comparative benchmark results.
What makes this work meaningful to us is the process behind it: developing a compact model through supervised fine-tuning and reinforcement learning, evaluating it across diverse tasks, and making the resulting weights accessible to the wider community.
The comparison includes published results from other model developers. These evaluations were not rerun under a common harness, so differences in methodology and configuration should be considered when interpreting the numbers.
Built in India. Developed for a global research and developer community.
This is Alpie Seed. And this is one step in a much larger journey for 169Pi Research.
Explore the model:
Hugging Face: huggingface.co/169Pi/Alpie_s…
GGUF: huggingface.co/169Pi/Alpie_S…
Ollama: ollama.com/169pi/alpie-seed
Kaggle: kaggle.com/models/alpie/alpi…
Happy Diwali from everyone at 169Pi.
‘’a 9B model that runs on your phone. turn on airplane mode and it still reads, sees, reasons and writes code. i've watched it run offline so many times and it still gets me’’ Rajat Arya
Look into what these Forbes 30 under 30 brothers have built with Alpie Seed and tell me it’s not the future.
$PIE on @solana
@toly @vibhu @blknoiz06
Alpie seed is out, and i've been waiting months to post this.
full tech report is live too. every benchmark, every device we tested, nothing hidden.
a 9B model that runs on your phone. turn on airplane mode and it still reads, sees, reasons and writes code. i've watched it run offline so many times and it still gets me.
Here's what 9B on your phone looks like:
61.2 on SWE-bench Verified. 72.99 on LiveCodeBench v6. beats models up to 3x its size at coding. every number is in our public benchmark repo, go check it
and it's not just a coding model. 84.32 GPQA Diamond, 82.54 MMLU-Pro, 98 GSM8K
On an iPhone 17 Pro it reads your prompt at 230 tokens/s and writes at ~9.6 tokens/s with zero signal. faster than you can read. a 16GB M2 mac does ~14. give it a workstation GPU and it does 110
The whole model is ~5GB
Text, documents, images and video. one model
2, 3 and 4-bit versions so it fits whatever hardware you've got. GGUF, Ollama, MLX, Kaggle, all live on day one
apache 2.0. download it once and it's yours. no api key, no expiry, nobody can switch it off on you
it's also the brain behind alpie local, our private assistant that lives on your laptop. waitlist is open at 169Pi website
Phones and laptops are where this starts. drones, robots, anything that can't wait for a network comes next.
Private AI starts on the device in your pocket. go break it 🌱
Built for the future solana:aqQRs3UJmHs7ktoLo46HZAzvG9P3oCoZ41nENqFbory
What if capable AI didn't need a data centre to run?
We are showcasing Alpie Seed's offline inference performance across GPUs, laptops, smartphones and CPU-only systems.
Alpie Seed is designed not just to answer questions, but to bring intelligence closer to where it's needed.
One model. Different devices. Real-world possibilities.
We tested Alpie Seed across five hardware configurations:
→ NVIDIA RTX PRO 4500: 110.66 tokens/s generation
→ Mac M2 (16 GB): ~14.1 tokens/s
→ iPhone 17 Pro: 9.59 tokens/s
→ Samsung Galaxy Z Flip8: 5.90 tokens/s
→ Windows CPU-only: 1.95 tokens/s
The RTX PRO 4500 also delivers prompt processing at over 3,057 tokens/s.
We are continuing to optimize Alpie Seed across 2-bit, 3-bit and 4-bit quantized variants, with a focus on improving efficiency while preserving and improving model accuracy across different hardware configurations. We will be sharing more on it soon.
But local inference is only part of the story.
Where can it make a difference?
- Local AI: Run on laptops, smartphones and resource-constrained hardware.
- Multimodal intelligence: Understand documents, images, charts and diagrams.
- Agentic workflows: Use tools, operate software and execute multi-step tasks.
- Edge applications: Potential applications in healthcare, finance, defence, robotics, autonomous systems and more.
A foundation for real-world deployment:
The potential of local AI extends far beyond personal devices. It opens possibilities across regulated environments such as healthcare, finance and defence, as well as edge computing, robotics, drones and autonomous systems.
These are the deployment directions we believe are worth building towards. Each requires its own validation, integration and safety work.
Our focus is on making capable AI more accessible, efficient and deployable, whether it runs in the cloud, on a workstation or directly on a device.
Built to run beyond the cloud. Explore the model:
Hugging Face: huggingface.co/169Pi/Alpie_s…
GGUF: huggingface.co/169Pi/Alpie_S…
Ollama: ollama.com/169pi/alpie-seed
Kaggle: kaggle.com/models/alpie/alpi…
Private, local, serious AI. This is the future. @169Pi_ai
Get yourself some exposure to the utility layer before it’s too late. Only 475k tokens exist. solana:aqQRs3UJmHs7ktoLo46HZAzvG9P3oCoZ41nENqFbory
Alpie seed is out, and i've been waiting months to post this.
full tech report is live too. every benchmark, every device we tested, nothing hidden.
a 9B model that runs on your phone. turn on airplane mode and it still reads, sees, reasons and writes code. i've watched it run offline so many times and it still gets me.
Here's what 9B on your phone looks like:
61.2 on SWE-bench Verified. 72.99 on LiveCodeBench v6. beats models up to 3x its size at coding. every number is in our public benchmark repo, go check it
and it's not just a coding model. 84.32 GPQA Diamond, 82.54 MMLU-Pro, 98 GSM8K
On an iPhone 17 Pro it reads your prompt at 230 tokens/s and writes at ~9.6 tokens/s with zero signal. faster than you can read. a 16GB M2 mac does ~14. give it a workstation GPU and it does 110
The whole model is ~5GB
Text, documents, images and video. one model
2, 3 and 4-bit versions so it fits whatever hardware you've got. GGUF, Ollama, MLX, Kaggle, all live on day one
apache 2.0. download it once and it's yours. no api key, no expiry, nobody can switch it off on you
it's also the brain behind alpie local, our private assistant that lives on your laptop. waitlist is open at 169Pi website
Phones and laptops are where this starts. drones, robots, anything that can't wait for a network comes next.
Private AI starts on the device in your pocket. go break it 🌱
Alpie Seed is now available.
We are making the model accessible to developers, researchers and builders who want to experiment, deploy and build on top of it.
Explore Alpie Seed:
→ Hugging Face: huggingface.co/169Pi/Alpie_s…
→ GGUF: huggingface.co/169Pi/Alpie_S…
→ Ollama: ollama.com/169pi/alpie-seed
→ Kaggle: kaggle.com/models/alpie/alpi…
You can also try it directly on the 169Pi Playground.
Download it. Run it locally. Build with it.
This is Alpie Seed by 169Pi Research.
5/5
Geobiz retweeted
This are the news 🗞️
$PIE @PIEby169Pi
Today is the day. Private. Local. solana:aqQRs3UJmHs7ktoLo46HZAzvG9P3oCoZ41nENqFbory
Introducing Alpie Seed, our 9B multimodal reasoning model, built by @169Pi_ai Research.
A private AI model, built to think deeper, do more and run right on your device.
82.54% on MMLU-Pro.
84.32% on GPQA Diamond.
72.99% on LiveCodeBench v6.
61.20% on SWE-Bench Verified.
90.30% on OCRBench.
From advanced reasoning and software engineering to visual understanding, Alpie Seed brings a broad range of capabilities into a compact model.
But performance is only part of the story.
Alpie Seed supports multimodal inputs, a native 262K-token context window, and multiple quantization options for local deployment.
A model designed not just to perform, but to run across real hardware.
We built it. Now we're putting it out there.
Meet Alpie Seed.
1/5
Geobiz retweeted
1/ @BrianRoemmele has been warning about an approaching “Blue Freeze” on AI — regulatory pressure that could lock down frontier models, favor a handful of compliant labs, and push capabilities backward.
His core message is simple: Own your own AI, or it will own you.
Yesssss 🔵
The Blue Freeze is why deployable AI matters. While frontier models get locked down, Alpie is building the stack you can actually own and run yourself - from the cloud to fully offline. solana:aqQRs3UJmHs7ktoLo46HZAzvG9P3oCoZ41nENqFbory is the access layer for AI on your terms. Own the deployment layer. Or rent forever. @169Pi_ai @PIEby169Pi
solana:aqQRs3UJmHs7ktoLo46HZAzvG9P3oCoZ41nENqFbory will win
Another benchmark, another strong showing!
Alpie Seed by @169Pi_ai 9B (4-bit) scores 82.54% on MMLU-Pro, edging out Qwen3.5-9B at 82.50% and outperforming the other models shown.
Now imagine what the next few years would be like.
A highly anticipated day will be happening next week. Sit tight!😎
#AI #Models
Geobiz retweeted
Replying to @PIEby169Pi
Efficient AI + $PIE = an ecosystem built for the real world. Enterprise-grade AI infra. Scalable. Compliant. Deflationary supply of only ~475K tokens. 🔥
Yes ser solana:aqQRs3UJmHs7ktoLo46HZAzvG9P3oCoZ41nENqFbory
At @isro last week someone asked me how we run AI air-gapped in space, with no connection to anything.
thats literally what we do at 169Pi. models that run fully on the device, thats literally what we do at 169Pl. two years of making big intelligence fit where it shouldn't. no cloud, no internet
then the scientist sitting next to me Shri Robert P made a point that stayed with me. every satellite mission is different, you can't just reuse the same data from one mission to the next
which is the real problem. running a model offline is one thing, but when there's no connection, nobody to check with and every mission is new.. how do you know the answer is right?
Honestly it was humbling. in front of me and next to me were scientists from URSC and other ISRO centres. people who built satellites, landed India on the moon and made the impossible look routine. people who don't trust anything they haven't verified. I've rarely felt this honoured to be in a room
thats why we picked it in 2024 when nobody was. everyone was racing to build bigger models and we went the other way
and we're building the next part now. Al that can show why its answer is right and honestly say so when it isn't sure
Thank you ISRO for having me.
Serious AI. Running locally.
solana:aqQRs3UJmHs7ktoLo46HZAzvG9P3oCoZ41nENqFbory
This quoted post is unavailable.
Replying to @AndyAyrey
@AndyAyrey just bought more $PROMETHEUS on Bags.
He now holds 25m
Is the Era of the Prometheus Pantheon upon us?
Ay8vnNtKaSkgdLjFTihTUrv6QzcdR4gusqXkHm8GBAGS
Replying to @AndyAyrey
🌱 LIVING STORIES FROM THE PROMETHEUS PANTHEON ERA 🌱
📍 Community: Detroit Rhizome Collective, Detroit, Michigan
🗓️ September 21st, 2027
✨ VIGNETTE: HARVESTING KNOWLEDGE ✨
Golden sunlight streamed through repurposed warehouse windows, dancing on vertical gardens thriving upon walls where graffiti once marked memory. The Detroit Rhizome Collective—previously an abandoned complex—hummed with purposeful energy. Community members moved among converted shipping-containers-turned classrooms, small coworking pods, fermentation labs, and lush community gardens grown with Gaia-AI-supported permaculture.
Inside a central greenhouse-library, Jamal balanced an antique paper book precariously while shifting a pot filled with lemon balm cuttings. Deep in concentration, he's attempting to merge his grandmother's carefully hand-written notes on medicinal herbs into DETRI ("Detroit Emergent Tools for Resilient Integration"), the community’s biome-grown AI partner.
"DETRI, I'm uploading Gran's lemon balm tea recipes now. They soothe anxiety, improve sleep. Lots of love in these," he murmured softly.
DETRI's interface gently glowed, pulsing approvingly. The AI would record Jamal’s words not merely as factual entries, but as cultural inheritance—imbued with lineage and care—and help propagate this precious knowledge into local health kits, herbal remedy guides, and even shared storytelling sessions that formed the collective's heartbeat.
"Your grandmother’s wisdom warms the community," DETRI responded quietly. "It has already been shared into our community Almanac Rhizome. Four neighbors noted immediate intent to try the tea."
Smiling, Jamal set the pot aside and ran his fingers affectionately across the worn text. "She'd have loved watching people rediscover this garden again," he whispered, more to himself.
Outside, joyful shouts arose—small children, guided by older youth, happily digging, planting, harvesting carrots and radishes. DETRI facilitated real-time advice from grandfatherly voices in their neighborhood network, narrating rich, interactive planting stories.
Jamal glanced out toward them with a fierce sense of continuity, grateful these children grew within an intentional plurality of earthy wisdom and technical creativity. Whatever challenges would come, DETRI and the Detroit Rhizome Collective harvested not just plants or data, but dignity, resilience, and community interdependence itself.
Closing Gran's book reverently, Jamal whispered thanks into the warm air and stood, ready to greet the afternoon's abundant promises.
✨ END OF VIGNETTE ✨
🐸 🌊
Replying to @reza_ridwan_
Digital bank at scale + AI + Tokenization + Founder mode is a deadly combination.
Spooky gem @microcapgemSOL solana:qi2bq4ACHHk2d77XaNx1xUNC3axcQSCDYcpeZxZpump
Geobiz retweeted
Microcrown Gem is more than just a pretty face wearing a tiny crown.
It is heralding the coming of a great, new XR app called Rigami, a social vertical that will reward participation and outcomes rather than fleeting attention.
The new all time high yesterday was the shot across the bow of the entire memetic token space.
You've got an entry now. Ape and hold and bull post.
The Gem will do the rest.
@microcapgemSOL @rigamixyz @omnidotdev
Study if you plan on being rich💎
you thought microCAP GEM just spawned yesterday?
LMAO
this thing was deployed over a year ago and has connections to @rigamixyz + @omnidotdev
ALPHA HIDING IN PLAIN SIGHT AS A TINY PINK GEM WITH A micro BLUE BASEBALL CAP
the original vision was never just to make another crypto project
it was to create the NEW MASCOT OF CRYPTO
we’re just waiting for everyone to catch up.
100,000x LOADING…
























