Pinned Tweet
You’re probably already overspending on AI, using frontier models for basic tasks like summarizing or classifying text.
By running smaller, routine AI tasks - summarization, classification, etc. - on the idle compute already sitting in phones, laptops and connected devices, we rgy demands for AI tasks by 400x compared to the same work done on a frontier model running in a data center.
By running smaller, routine AI tasks - summarization, classification, etc. - on the idle compute already sitting in phones, laptops and connected devices, we rgy demands for AI tasks by 400x compared to the same work done on a frontier model running in a data center.
3
4
8
73,542
1,900x less energy than a data center.
In our edge network benchmark, 64 million AI requests on idle devices used about 8 kWh. A data center would have used 15,353 kWh for the same requests.
zerogpu.ai/benchmarks/edg…
In our edge network benchmark, 64 million AI requests on idle devices used about 8 kWh. A data center would have used 15,353 kWh for the same requests.
zerogpu.ai/benchmarks/edg…
0
0
0
10
15,345 kWh less electricity than a data center would use, in 48 hours. That's more than the average home uses in a year!
From our edge network benchmark: 64 million AI requests on idle devices, about 8 kWh in total.
zerogpu.ai/benchmarks/edg…
From our edge network benchmark: 64 million AI requests on idle devices, about 8 kWh in total.
zerogpu.ai/benchmarks/edg…
0
1
2
37
Our small language model for summarization, llama-3.1-8b-instruct-fast, reads a whole call or email thread in one pass with a 128K token context window. No chunking.
Built for sales intelligence, contact centers and meeting notes.
docs.zerogpu.ai/api-reference/…
Built for sales intelligence, contact centers and meeting notes.
docs.zerogpu.ai/api-reference/…
0
0
1
20
ChatGPT wants maximum utilization in the cloud. We'd rather use the idle compute people already own, so we pay device owners to put it to work.
OpenAI CEO Sam Altman believes people owning RAM for their own PCs is a waste and instead should be saved for data centers
"It’s quite terrible to have everyone have a bunch of the world’s limited supply of DRAM sitting in their own computer."
0
0
0
23
OpenAI wants maximum utilization in the cloud. We'd rather use the idle compute people already own, so we pay device owners to put it to work.
OpenAI CEO Sam Altman believes people owning RAM is wasteful, and that it belongs in data centers instead.
"From a global allocation perspective, actually, it’s quite terrible to have everyone have a bunch of the world’s limited supply of DRAM sitting in their own computer."
0
0
0
19
Turn a request into a tool call. Our ZLM for tool decisions, zlm-v1-tool-decisions, tells your agent which MCP tool to call and what to send it, and returns NO_TOOL when none is needed.
docs.zerogpu.ai/api-reference/…
docs.zerogpu.ai/api-reference/…
0
0
0
24
ZeroGPU AI retweeted
We gave Claude Code one prompt to classify 20 websites with @ZeroGPU_AI. It chose the right model, paid and ran all 20 calls through ZeroClick, with no help after the prompt.
Here's the demo and what we learned:
zeroclick.ai/blog/how-zerog…
Here's the demo and what we learned:
zeroclick.ai/blog/how-zerog…
0
3
12
306
ChatGPT wants maximum utilization in the cloud. We'd rather use the idle compute people already own, so we pay device owners to put it to work.
OpenAI CEO Sam Altman believes people owning RAM for their own PCs is a waste and instead should be saved for data centers
"It’s quite terrible to have everyone have a bunch of the world’s limited supply of DRAM sitting in their own computer."
0
0
0
15
OpenAI wants maximum utilization in the cloud. We'd rather use the idle compute people already own, so we pay device owners to put it to work.
OpenAI CEO Sam Altman believes people owning RAM is wasteful, and that it belongs in data centers instead.
"From a global allocation perspective, actually, it’s quite terrible to have everyone have a bunch of the world’s limited supply of DRAM sitting in their own computer."
0
0
0
17
Our small language model for entity extraction, gliner2.5-base-v1, pulls people, companies and products out of any text with your own labels. Input tokens cost 10x less than GPT-5.4 nano.
Built for financial research, news monitoring and search.
zerogpu.ai/use-cases/sent…
Built for financial research, news monitoring and search.
zerogpu.ai/use-cases/sent…
0
1
1
38
AI inference. Closer to you.
CDNs bring content closer to users. ZeroGPU brings AI inference closer to users.
Geo-aware routing is live:
🌍 200,000+ devices
📍 140+ countries
⚡ 20% lower latency with geo-aware routing
CDNs bring content closer to users. ZeroGPU brings AI inference closer to users.
Geo-aware routing is live:
🌍 200,000+ devices
📍 140+ countries
⚡ 20% lower latency with geo-aware routing
1
0
0
17
Our mission: orchestrating the world's idle compute to accelerate AI.
Geo-aware routing sends supported workloads to nearby available devices, so idle compute works closer to demand.
Try it with your workload: zerogpu.ai
Geo-aware routing sends supported workloads to nearby available devices, so idle compute works closer to demand.
Try it with your workload: zerogpu.ai
0
0
0
7
Set a schema, get JSON back. Our small language model for extraction, gliner2.5-base-v1, pulls exact fields out of messy documents, at $0.02 per 1M input tokens.
Built for recruiting, insurance and lending.
docs.zerogpu.ai/cookbook/data-…
Built for recruiting, insurance and lending.
docs.zerogpu.ai/cookbook/data-…
0
0
1
46
"These small models are going to get more and more important as compute gets tighter and tighter."
Heard on All-In this week. That is the bet we built ZeroGPU on: small and open-weight models, run where the compute already is.
Heard on All-In this week. That is the bet we built ZeroGPU on: small and open-weight models, run where the compute already is.
0
0
0
35
Our audio models now support batch tasks. Queue a backlog of recordings for whisper-tiny speech to text, or a list of scripts for chatterbox-nano text to speech, and collect the results when the job finishes. docs.zerogpu.ai/api-reference/…
0
0
0
33
29:20 - Perplexity CEO @aravindsrinivas says "the data center is coming to your laptop."
That’s what we’re building.
We're already driving major savings for AI users, who are paying 50% to 90% less for inference.
That’s what we’re building.
We're already driving major savings for AI users, who are paying 50% to 90% less for inference.
NEW: I sat down w/ Perplexity Founder & CEO Aravind Srinivas (@AravSrinivas) on why AI-native SMBs could drive the next American millionaire economic boom.
@perplexity_ai, valued at ~$20B with ~$2B in reported funding, $750M in annualized revenue as of August, seeing 5x YoY reves from search into autonomous agents.
One example Aravind shared is a single-person company generating $3M in annual revenue while spending over $1M a year on Perplexity Computer.
His thesis is that AI could fundamentally change the economics of building a business. Smaller teams, less capital, higher productivity, & entirely new business models.
At the infrastructure layer, Aravind sees a shift toward local & hybrid inference, with AI agents operating across personal devices and the cloud. As context windows expand and workloads become more personalized, local compute becomes increasingly cost-competitive, while offering greater privacy, control & data sovereignty.
We cover:
› Why SMBs could become a major driver of US GDP growth
› The shift from AI models to agents & recursive economics
› Why Perplexity is betting on your Mac instead of building data centers
› NVIDIA, Apple Silicon, local inference & sovereign AI
› Why writing code is no longer the bottleneck, but problem-solving is
𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒
(00:00) Aravind Srinivas, Co-founder & CEO of Perplexity
(01:30) Why Perplexity's brand is built on curiosity
(04:47) Every great invention starts with one question
(08:18) How to build a $1T company in America alone
(12:36) The next big companies will be tiny
(13:45) Is the American dream still alive?
(16:28) Stop blaming users for bad prompts
(22:55) How Perplexity quietly grew 5x
(26:51) The solopreneur spending $1M a year on Perplexity
(28:28) "We don't have competitors, we have contributors"
(29:20) The data center is coming to your laptop
(34:32) Orchestrating up with NVIDIA, down with Apple
(36:11) What happens when your AI agent knows everything
(39:17) Why Perplexity is skipping the data center race
(41:25) Stop asking AI to write code, ask it to think
(45:47) Perplexity’s answer to AI security
(48:27) The Hugging Face incident & multi-agent threats
(50:45) Turning everyone into a scientist
(53:31) Why Perplexity isn’t chasing a billion users
1
0
0
45
Running AI inference through ZeroGPU takes a fraction of the energy compared to a frontier model running on a data center.
In fact, we use up to 1,900× less per request in our latest 48-hour test.
zerogpu.ai/benchmarks/edg…
In fact, we use up to 1,900× less per request in our latest 48-hour test.
zerogpu.ai/benchmarks/edg…
0
0
0
17