Jatevo
3K posts
@JatevoId
Infrastructure for Private AI. Fast LLM Inference powering AI agents via x402. CA: 9VY2rDbtsBmTsBxoRF8hWSEUKGqnoQoe9V6W3JnjNgfm
In the last two weeks, we processed more than 20 billion GPT‑5.5 tokens through $JTVO inference. A wallet holding $2,000 worth of $JTVO (eligible for 2,000 API requests per day) spent more than $5,000 on GPT‑5.5 API requests. We are actively increasing nodes to improve performance and lowering the latency. Stats below👇🏻
Today on MCG @JatevoId w/@lucaxyzz | solana:9VY2rDbtsBmTsBxoRF8hWSEUKGqnoQoe9V6W3JnjNgfm Jatevo is building the infrastructure for private AI Luca joins back on the show to drop major updates and tease what's coming down the line Highlights: 02:57 - Was talking decentralized inference in early ICM before anyone understood it 04:24 - Major B2B compute deals 06:35 - Luca's biggest mistake 09:42 - One API key, eight models now, more than ten coming 10:49 - How to win enterprise buyers 13:03 - GPU prices movement 21:10 - Cost to host a model 22:35 - @BytePlusGlobal partnership 27:57 - Key metrics
People will rush to get a DGX Spark. Soon, they'll be able to rent it easily via jatevo.ai/decentralized. Community-powered inference.
Day-3 warming up the DGX
used on an agentic harness so far. Gemma 4 31B is rapid! Pretty instant response but needs detailed prompts to actually follow. Agentic tool calling needs some work and skills need reminders, but it's been okay considering the speed. Generates basic HTML pages in seconds, so good for quick info dumps and revision guides. Haven't benched it or started a real build, but for speed and the basics, it's solid. @JatevoId and @lucaxyzz were kind enough to chuck me some credits to keep the addiction at bay.
used on an agentic harness so far. Gemma 4 31B is rapid! Pretty instant response but needs detailed prompts to actually follow. Agentic tool calling needs some work and skills need reminders, but it's been okay considering the speed. Generates basic HTML pages in seconds, so good for quick info dumps and revision guides. Haven't benched it or started a real build, but for speed and the basics, it's solid. @JatevoId and @lucaxyzz were kind enough to chuck me some credits to keep the addiction at bay.
Gemma 4 31B via @cerebras is live at jatevo.ai/playground up to 3000 token/sec. API access will be available for $jtvo holders soon!
Gemma 4 31B via @cerebras is live at jatevo.ai/playground up to 3000 token/sec. API access will be available for $jtvo holders soon!
In the last two weeks, we processed more than 20 billion GPT‑5.5 tokens through $JTVO inference. A wallet holding $2,000 worth of $JTVO (eligible for 2,000 API requests per day) spent more than $5,000 on GPT‑5.5 API requests. We are actively increasing nodes to improve performance and lowering the latency. Stats below👇🏻
Check your jatevo.ai dashboard, you need to hold a minimum of 1M $JTVO tokens to be eligible to access $POD @dphnAI. Our current infrastructure supports 120 RPM for all users. We're working on adding more GPUs to host this model if demand grows.
USERS REPORT ISSUES AT OPENAI, CLAUDE - DOWNDETECTOR
