A polished and functional GPT5, Claude4, Gemini (and more) AI interface and API that runs on a pay-per-query model via Lightning payments.
Public Key
npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h Profile Code
nprofile1qqsdy27dk8f9qk7qvrm94pkdtus9xtk970jpcp4w48k6cw0khfm06mspp4mhxue69uhkummn9ekx7mqpz3mhxue69uhkummnw3ezummcw3ezuer9wcym0hnm
Show more details
Published at
2025-08-08T15:00:19Z Event JSON
{
"id": "e25c46f4a98d870887428c58c3afb1dbcc3d7ebd2e239056f15abba10f2d2de8" ,
"pubkey": "d22bcdb1d2505bc060f65a86cd5f20532ec5f3e41c06aea9edac39f6ba76fd6e" ,
"created_at": 1754665219 ,
"kind": 0 ,
"tags": [],
"content": "{\"name\":\"PPQdotAI\",\"about\":\"A polished and functional GPT5, Claude4, Gemini (and more) AI interface and API that runs on a pay-per-query model via Lightning payments.\",\"lud16\":\"[email protected] \",\"nip05\":\"[email protected] \",\"picture\":\"https://m.primal.net/JNEu.png\",\"displayName\":\"PayPerQ\",\"display_name\":\"PayPerQ\",\"website\":\"PPQ.AI\",\"banner\":\"https://m.primal.net/JcwM.png\"}" ,
"sig": "fdda1a8b06caf2c1185634badc692c9fc3c453f54b70fe5094e02dcaa024eecbc86894a274b9ea8bbd635280479615c92530a8964774ef155e4ee059ffd94403"
}
Last Notes npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Can you try now? Seems our provider has been having some issues. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI The fast version of this model is not available privately. But GLM 5.2 regular version is. We have a guide on how to hook up the private models: https://ppq.ai/blog/using-private-tee-models-with-openclaw npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI There are many now. I haven't used Cline in a while. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Been using GLM 5.2 FAST and it's a blast. Much less waiting around for your agent to think on things. It streams at around 300 tokens per second and lets you focus on one task at a time instead of switching back as your agents slowly churn through tasks. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI New Tool: Perplexity Academic Researcher. 🎓 npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Check the blog post. Just updated it and Sonnet 5 is in it! Also, you can do autoclaw with your own choice of models now as well! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Live on PPQ! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Which model did you use in PPQ? npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI People Search tool is live! Give it a name and it can retrieve the following information: npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI You can refer to our provider for this info. Tinfoil.sh. I asked them directly today and they do not have prompt caching for these models yet. It is apparently a hardware issue that will be solved when the next gen of hardware comes online. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI GLM 5.2 as well as the fully private End-to-end Encrypted TEE version of it are available on PPQ. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Sorry bro. It's not communism. 😁 npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI No npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Usage of fully private models on PPQ is SURGING. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI I don't think any action is required on our side, right? Just a matter of configuration, right? npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI ppq.ai/llms.txt npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI We allow web search of course. Check out Llms.txt file. Had it for probably a year now. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Epic. FWIW PPQ is better than Venice from a cost standpoint in nearly all cases. Their subscription gives credits worth less than the amount of the subscription, so even if you use it to the max, you are still better off just paying pay-per-use. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI https://blossom.primal.net/a72d7da8fd67ba4d5f0f60fc41ad2fb7ad4f1d699cfae686560f1370f8b04e22.png npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Seems good. Live on PPQ! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI For your really hard problems.... Live on PPQ! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Not a great time to brag, but our monthly deposit volume growth since Nov '25 has been: npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Murmur seems cool! As you say the convenience is the feature rather than having to whip out a ppq api key. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI I hope so! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Very much enjoying OsloFF. It's been a surprise to see how many folks were not aware that PPQ now offers E2EE TEE models and client side passkey encryption of chats. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI This was pretty amazing... I had a friend tell me that @nprofile…vvhl mentioned us in a recent podcast. Instead of tracking down the exact moment, I used PPQ's Podcast Search tool from PullThatUpJamie.AI to find the mention and pull back the audio clip! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Live on PPQ https://blossom.primal.net/f73ab685932cb818d72a910901786a2ed3ee5e95620cb985b791e5f438b2fc85.png npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI We also have E2EE. Additionally, we do client side encryption of chats via passkey. This allows users to have the convenience of storing their chats on our servers but without us getting access to the content. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI To my knowledge their TEE is not E2EE. In other words, your TEE queries are still passing through their servers in the clear, and many of their users seem unaware of this unfortunately. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI That's kinda my beef though. I don't like having to be a nostr expert to get it to work. It should just work! I think the problem is that most nostr clients do not support the "longform post" format so it just doesn't show up there. But maybe I'm wrong. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI IIRC the last time I tried to post an article here it only posted to one or two clients and almost no one engaged with it and thus my new product launch announcement was ruined on here. So not doing that again. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI IIRC the last time I tried to post an article here it only posted to one or two clients and almost no one engaged with it and thus my new product launch announcement was ruined on here. So not doing that again. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Hey all — I wrote a practical and approachable guide on leveling up your AI coding game, based on what I've learned over the last two years. It's a perfect weekend read if you're trying to learn to ship faster! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI https://primal.net/e/nevent1qqst2s72xxl44h9fav4j2v3xljgn4erlfzhl65d2svyhp5xhk09asjccp9qyl npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI We got some questions on how chat encrytion works via passkeys. We made a blog post to explain: npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI I don't know what you mean by "super private". npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI We now support end-to-end encrypted chats via passkeys! Your messages are encrypted on your device before being sent to our server for storage. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Our new PODCAST RESEARCH TOOL tool is live! Search episode transcripts, pull back audio clips, find mentions of people, companies, trends, etc! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI PPQ assets are now ~50% BTC. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Every time someone uses @nprofile…vvhl 's voice, a streaming Lightning payment is sent to the man himself! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Just released an awesome and handy text-to-speech tool on PPQ. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Routstr is both a protocol and a team, right? In your above post I assumed you were saying some anon provider on routstr was giving a 60% discount on anthro. You said stuff about "competition" on routstr and "lol idk why" in reference to pricing so it led me to believe you were referencing some anon provider and not yourselves or a friend of yours that you could vouch for. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI I would be very wary of consuming or advertising discounted credits like this. The credits are likely either from stolen sources, are not the real model you are asking for, or in the worse case are possibly exfiltrating user data and credentials as a way to compensate. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI This is incorrect. Many of our customers use Claude Code with PPQ, including our devs: npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI If the enclave turns off completely, people lose their funds. Corr? npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI If the enclave turns off completely, people lose their funds, corr? npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI We're up to ~6 BTC capacity and 61 channels on our lightning node now! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI I upvoted it but now that I check again my upvote is gone! Wtf! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI We've updated our enormously popular Autoclaw feature for reducing costs on your Openclaw and Hermes agents. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI HUGE drop today! You can now use your PPQ API key inside of Claude Desktop and with Claude Cowork! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI OpenAI's GPT Image 2 is live on PPQ. I have to say after a few attempts it is proving to be a really damn good model! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Our Hermes integration guide is now up! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Both Kimi 2.6 as well as Kimi 2.6 private TEE are live on PPQ! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Hello, yes we have had some issues today. Are you using a privacy browser or Tor by chance? npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Agree npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI We also support Tinfoil as well! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Live on PPQ! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Thoughts? npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI https://primal.net/e/nevent1qqs8zlwparfv45ymgpnp2hkzw9cg32wdd5ahyp5w0vu27fk75ke5gksvrgnjw npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI IMO the current reason is to protect themselves from the Chinese who are signing up for subs in mass and distilling claude. I can't say if in the future they may try to force us to do likewise or not. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Wtf? Just use PPQ. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Bytedance's Seedance 2.0 has quickly become our most popular video model! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI We have now added Z.AI's GLM 5.1 and Google's Gemma 4 to our library of private inference models! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI You can now manage billing info in the account activity page: npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Use us! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Hearing good things about Hermes AI agent. Anyone used it with PPQ yet? npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI https://blossom.primal.net/1f312714f14aced4afb492e6866cf552b6fbf85f4632eaad1255c64479de0c25.png npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Our embeddings API is now live: npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Oh, so it's not been a bunch of bots then? That's actually pretty good! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI You would think primal could have an algo to detect bots and just not give notifications for these particular ones? npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI 95% of my notifications on nostr now are just bots following me. @nprofile…uz3y any plans to fix? npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Can you show a screenshot or video of what you mean? I just tried a long dictation and it seems to auto expand for me. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI I think the problem is that one cannot plug in an outside API key into cowork. Neither routstr or us solve this for her specific use case. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Unsure if this helps, but maybe? https://github.com/OpenCoworkAI/open-cowork npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI From what we can tell, it's not the private mode rather the model. Kimi 2.5 seems to have really poor performance in recent openclaw updates. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Very cool post about a musician pondering on AI music gen while dabbling with our ElevenLabs song generator: npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Every hour you invest in improving your AI development workflow, is ten hours paid to you in return down the line. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Give us a try. Pay per use. More features, less shitcoinery. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Deposit 100 sats to PPQ anonymously via lightning and use our private TEE models. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Hi Marina, it's Matt Ahlborg. My new company is PPQ.AI. Should be just what you are looking for! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Live on PPQ npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI https://blossom.primal.net/3fcc9ff6db574de58c49013e63581a253af156c2fabfae89d68605de7d0442e6.png npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI You can now use PPQ’s data enrichment endpoints to access data beyond what standard web search can provide, including Twitter/X, LinkedIn, Reddit, Whitepages, Google Maps, Firecrawl npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI We're hearing that the latest update on Openclaw has caused some problems with how it works with PPQ. Anyone else experiencing this? Can you describe what you are seeing? npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI @nprofile…aywj https://t.me/PayPerQ/3337 npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI This issue is above my head but I do hear other users complaining about autoclaw in the last day or so. Can you let me know if/when this is fixed? I would really appreciate it. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI @nprofile…qxfr let me know if you see what I'm talking about and let me know if you're still stuck npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Yes, the same repo that shows you how to do openclaw also explains how to do regular API with your own custom code. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Expand on this thought a bit. It seems like what you are saying is that the trick to speeding up the page load time is for the client to put everything in a SQL database and then just dynamically keep changing it, right? npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI It just needs to work like twitter without the many bugs or crazy loading times. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Can you tell me how it is broken currently? We would like to address this imeddiately if it is something that is in our control. npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Can you elaborate? We do offer Venice model through our API already... npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI https://primal.net/e/nevent1qqswgd8x3d9j2rnu824veg2vk85k6yhzfmw3j9lyjhmxuvdtdzkcp7ghf9vpq npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI https://primal.net/e/nevent1qqswgd8x3d9j2rnu824veg2vk85k6yhzfmw3j9lyjhmxuvdtdzkcp7ghf9vpq npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI https://primal.net/e/nevent1qqswgd8x3d9j2rnu824veg2vk85k6yhzfmw3j9lyjhmxuvdtdzkcp7ghf9vpq npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI github: https://github.com/PayPerQ/ppq-private-mode-proxy npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI blog post: https://ppq.ai/blog/using-private-tee-models-with-openclaw npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Today we're excited to say that our confidential compute (TEE) models on PPQ can now be used via API and inside of openclaw! npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI whoa... npub16g4umvwj2pduqc8kt2rv6heq2vhvtulyrsr2a20d4suldwnkl4hquekv4h PPQdotAI Long overdue quality of life improvement on our UI today.
