AI is getting more interesting, and it's not just about building bigger models anymore.
Coinbase recently fine-tuned Qwen3.5-9B using internal fraud data, reportedly outperforming its Claude Opus 4.5-based reviewer across four fraud metrics. The surprising part? It cost less than $100.
Now here's where Acurast comes in.
The same AI model can run on a 16 GB Android smartphone, with reported operating costs of just $0.05–$0.10 per day.
Think about that. A regular smartphone potentially becoming part of the infrastructure powering AI.
To me, this is where things get exciting.
Smaller, specialized AI models combined with decentralized computing could reduce costs and make AI infrastructure accessible to more people.
Maybe the future of AI isn't just bigger data centers. Maybe it's millions of devices working together.
Definitely keeping an eye on
@Acurast $ACU
Coinbase recently fine-tuned Qwen3.5-9B using internal fraud data, reportedly outperforming its Claude Opus 4.5-based reviewer across four fraud metrics. The surprising part? It cost less than $100.
Now here's where Acurast comes in.
The same AI model can run on a 16 GB Android smartphone, with reported operating costs of just $0.05–$0.10 per day.
Think about that. A regular smartphone potentially becoming part of the infrastructure powering AI.
To me, this is where things get exciting.
Smaller, specialized AI models combined with decentralized computing could reduce costs and make AI infrastructure accessible to more people.
Maybe the future of AI isn't just bigger data centers. Maybe it's millions of devices working together.
Definitely keeping an eye on
@Acurast $ACU
