@drae_drae22@puppyBulldogFR@169Pi_ai Check out post from @V2Chenz above - all you need to know β¦ Tech is A grade and they have huge real world partnerships in the works. Working on token integration .. millions, easy peasy π
@BizChirag his first TED Talk is now live on their official YouTube channel with 44 million subscribers.
Additionally, we have begun an organic marketing plan for Alpie, and you will see the implementation soon π
Link:
https://t.co/N2Mk0u7zRt
A user opened ALPIE CORE, typed a few prompts, and walked away with 15 fully built projects.
The bill? $0.017.
We saw it first on X and honestly, we had to sit with it for a moment. Not because it surprised us - but because it was the clearest real-world proof yet of what we've been quietly building toward.
Here's what makes this moment worth understanding properly.
ALPIE CORE is our 4-bit 32 billion reasoning model built specifically for real-world deployment. Not a lab experiment. Not a full-precision benchmark setup. A model designed to actually run efficiently at scale, in the hands of real users, at real costs.
And on SWE-Bench Verified - one of the most rigorous software engineering benchmarks that exists - it scored 57.8.
The best-performing model at the 32B quantised scale. Built to be deployed. Built to be affordable. Built to work.
So when that user built 15 projects for $0.017, it wasn't luck. It was the direct result of a model that was deliberately engineered to be powerful AND practical at the same time.
Most capable models are expensive to run. Most cheap models aren't capable enough. We've been obsessing over closing that gap.
This is what closing that gap looks like.
We are continuously improving ALPIE CORE, pushing its limits with every iteration. ALPIE CORE V2 is already in the works and if this is what V1 looks like, we can't wait to show you what's coming next.
What will you build with @169Pi_ai ?
$PIE @ $442K has 1000x potential. Let me give you details as to why.
Alpie Core is a frontier 32B 4-bit quantized reasoning model from @169Pi_ai ... India's first at this scale rivaling or beating GPT-4o/Claude 3.5 Sonnet on reasoning, coding (57.8% SWE-Bench), and math benchmarks while using a whopping 75% less memory, 3x faster inference, and 10x cheaper. Trained on 8 H100 GPU.
Alpie playground powers production AI agents via its unified reasoning API that is already driving real-world gains like,
- Deep Research Agent: +34% accuracy, 3.2x faster synthesis
-PDF Analysis Agent: 92% legal docs, 87% scientific papers, 91% financial reports
Runs 3x faster, 10x cheaper than frontier models, deployable on consumer GPUs 8x H100. No massive infra needed. Ideal base layer for scalable, sustainable agent economies on-chain or off.
And their native platform utility token powers this with priority access, incentives, platform credits, discounted inference, unlock workspace features, extended memory, model labs and much more.
The team are doxxed. Rajat worked at Netflix and Snapchat then left and founded Spheric; Spheric was a blockchain-focused tech services company he co-founded (around 2020β2023 timeframe), specializing in delivering tech, product, and infrastructure solutions, particularly for clients in North America and the Middle East.
He was also involved with building LotanChain Achieving up to 65,000 TPS.
Both the founders have been features in Forbes 30 U 30 as the only AI start up.
Are media partners with NDTV where they were present at Davos and India's largest AI summit as a speaker.
Are partnered with ISRO (Equivalent of NASA in India) a strategic collaboration focused on deploying and advancing sovereign, efficient AI systems for India's national missions, particularly in secure, data-sensitive environments like space tech.
Rajat has hosted talks on TedXKiet about AI advancement.
There is a plethora of other catalysts.
The token has real world utility, what was speculative drive PA will soon be utility driven > https://t.co/FSdCrqOK22
> https://t.co/qXbpJTnFNs
> https://t.co/BlNI41Zml0
I know the markets aren't favourable.
But they are in talks with government agencies, running pilots with US enterprise and have been labelled top 5 Indian AI start ups by Inc42.
The downside is fkn minimal. The upside is absolutely massive.
A reasoning model that beats Qwen, OpenAI and several other legacy models in cost and efficiency, but also reasoning.
Not to mention, India is the fastest growing major economy and is top 3 within the AI race.
1/ While Sarvam & Krutrim chase billions in funding for bloated models, the Arya brothers (@rajatarya01 & @BizChirag) are quietly executing at https://t.co/kIi2wyghKV.
Compare the tape
Krutrim - $1B+ unicorn, $300M+ raised
Sarvam - $41-53M raised
these are some of India's leading AI start ups, all of which are funded by venture capital. Alpie Core is not.
- Neysa: $1.4B - In development since 2023
- @OfficialKoreAI: $840M - In development since 2013
- @qure_ai: $264M - In development since 2016
- @yellowdotai: $422M - In development since 2016
- @uniphore: $2.5B - In development since 2008
- @169Pi_ai: $350K - In development since late 2024.
$PIE @ $442K has 1000x potential. Let me give you details as to why.
Alpie Core is a frontier 32B 4-bit quantized reasoning model from @169Pi_ai ... India's first at this scale rivaling or beating GPT-4o/Claude 3.5 Sonnet on reasoning, coding (57.8% SWE-Bench), and math benchmarks while using a whopping 75% less memory, 3x faster inference, and 10x cheaper. Trained on 8 H100 GPU.
Alpie playground powers production AI agents via its unified reasoning API that is already driving real-world gains like,
- Deep Research Agent: +34% accuracy, 3.2x faster synthesis
-PDF Analysis Agent: 92% legal docs, 87% scientific papers, 91% financial reports
Runs 3x faster, 10x cheaper than frontier models, deployable on consumer GPUs 8x H100. No massive infra needed. Ideal base layer for scalable, sustainable agent economies on-chain or off.
And their native platform utility token powers this with priority access, incentives, platform credits, discounted inference, unlock workspace features, extended memory, model labs and much more.
The team are doxxed. Rajat worked at Netflix and Snapchat then left and founded Spheric; Spheric was a blockchain-focused tech services company he co-founded (around 2020β2023 timeframe), specializing in delivering tech, product, and infrastructure solutions, particularly for clients in North America and the Middle East.
He was also involved with building LotanChain Achieving up to 65,000 TPS.
Both the founders have been features in Forbes 30 U 30 as the only AI start up.
Are media partners with NDTV where they were present at Davos and India's largest AI summit as a speaker.
Are partnered with ISRO (Equivalent of NASA in India) a strategic collaboration focused on deploying and advancing sovereign, efficient AI systems for India's national missions, particularly in secure, data-sensitive environments like space tech.
Rajat has hosted talks on TedXKiet about AI advancement.
There is a plethora of other catalysts.
The token has real world utility, what was speculative drive PA will soon be utility driven > https://t.co/FSdCrqOK22
> https://t.co/qXbpJTnFNs
> https://t.co/BlNI41Zml0
I know the markets aren't favourable.
But they are in talks with government agencies, running pilots with US enterprise and have been labelled top 5 Indian AI start ups by Inc42.
The downside is fkn minimal. The upside is absolutely massive.
A reasoning model that beats Qwen, OpenAI and several other legacy models in cost and efficiency, but also reasoning.
Not to mention, India is the fastest growing major economy and is top 3 within the AI race.
1/ While Sarvam & Krutrim chase billions in funding for bloated models, the Arya brothers (@rajatarya01 & @BizChirag) are quietly executing at https://t.co/kIi2wyghKV.
Compare the tape
Krutrim - $1B+ unicorn, $300M+ raised
Sarvam - $41-53M raised
these are some of India's leading AI start ups, all of which are funded by venture capital. Alpie Core is not.
- Neysa: $1.4B - In development since 2023
- @OfficialKoreAI: $840M - In development since 2013
- @qure_ai: $264M - In development since 2016
- @yellowdotai: $422M - In development since 2016
- @uniphore: $2.5B - In development since 2008
- @169Pi_ai: $350K - In development since late 2024.