We’ve been working to make powerful AI accessible to everyone through our free product and low-cost subscription tier, ChatGPT Go—which is now available globally in every country where ChatGPT is available.
In the coming weeks, we’re also planning to start testing ads in the U.S. for the free and Go tiers, so more people can benefit from our tools with fewer usage limits or without having to pay. Plus, Pro, Business, and Enterprise subscriptions will not include ads.
We’re sharing our principles early on how we’ll approach ads–guided by putting user trust and transparency first as we work to make AI accessible to everyone.
What matters most:
- Responses in ChatGPT will not be influenced by ads.
- Ads are always separate and clearly labeled.
- Your conversations are private from advertisers.
- Plus, Pro, Business, and Enterprise tiers will not have ads.
Once we begin testing our first ad formats, we look forward to getting your feedback and ensuring that ads can support broad access to AI and keep the trust that makes ChatGPT valuable.
Did you expect to have trillion companies burning trees and paying for power plants for the free? It is only the beginning and will become very bad very soon.
Was meant to end up like that from the start.
Also the whole point of AI browsers + smart glasses, shoving MOAR ads into your retinas that know how you feel and what you like!
My biggest concern is that they’re going to start promoting semi-false recommendations or answers based on who pays more.
For example:
A user asks what’s the best cloud service and it says Google Drive, not because it’s the best option for that person, but because Google is paying OpenAI to recommend their service.
Or maybe it won’t explicitly say “Google Drive is the best,” but it will make it look like the best option even when it isn’t.
Is there any real perk to using the big tech LLMs? Like at all? I haven’t played with any of the big ones except for Copilot and I havent even touched that since early 2024 and have been playing with Ollama and Proton Lumo and I dont feel like im missing out on anything crazy. Normally the idea that we sell our data is exchange for something we can’t afford normally and of a certain quality of life hack but Im not even seeing that appeal here. Only thing that its got going for it is that it strokes your ego which as someone who desires accountibility in my life. That is not appealing to me in the slightest especially since this is not even coming from other humans. Ive set boundary that my conversations with AI will always remain ephemeral where possible. Past conversations would ruin any sense of objectivity. This feels icky knowing the types of sensitive conversations I can already forsee how this can be dangerous. I look at this as no different than being reccomended to buy crap from the tiktok shop or sponsored content on YouTube. Yeah there is supposed to protections in place but how often are they actually enforced?! I have 0 trust in any of the Big Tech to actually give a thought to this I think they just wanna ease into it that way we are numb when the worst finally happens I can already forsee all the issues with this. Plus if we have learned anything from Meta with messengers is that the content of the conversations itself is not needed for it to be invasive.
Lumo is trash tier LLM in comparison to big tech LLMs. It runs: “Nemo, OpenHands 32B, OLMO 2 32B, GPT-OSS 120B, Qwen, Ernie 4.5 VL 28B, Apertus, and Kimi K2” which are not super impressive. Ollama is only as good as your hardware which can be very expensive.
I was unimpressed with anything from OpenAI. I think Google’s models are top notch and I really like their gemma 3 abliterated 28B model for local fun. For work (code) I believe Anthropic is unrivaled. Sonnet 4.5 is pretty good for most things, but Opus 4.5 is the deep pockets, 3x the Sonnet price model that I consider state of the art (objective - no way to truly measure).
Im mainly have 2 use cases for Lumo. 1. To assist me in mixing within DAWs when Im looking to make a specific type of sound with the intention of still following my ears and other peoples input 2. to find sources for claims that I have not found by a typical search.
I dont have crazy requirements with Ollama either Its basically as a doomsday prep to use alongside a Kiwix database to still have some accessibility of knowledge when internet is gone that is meant to be applied in the real world. Discernment is key for this. Im using mistral:7b and llama3.1:8b which in my admittedly brief testing analyzes a few wiki articles just fine
I like Lumo. Proton are still adding features and it has web search functionality. I use it instead of a browser most of the time, especially on devices I don’t trust or control. The free option works without signing in. I have a paid subscription and feel it offers reasonable value for money as an existing Proton customer. It also integrates with Proton Drive. They are adding the ability to generate and analyse images. If the price doesn’t increase that is good value.
Lumo is censored but explaining why I’m asking usually helps. Self hosting is better in that regard but Lumo rarely refuses to answer a question after enough coaxing. Factual accuracy seems acceptable. Sometimes it contradicts itself from one conversation to another but that is to be expected.
My experience of the competition is limited so maybe my standards are lower. It is good enough for general problem solving, document analysis, and augmenting my day to day life. It doesn’t affirm everything I say and gives a balanced viewpoint. That may be due to the way I set it up.
I like how it doesn’t try to get to know me and acts as an assistant, not a friend. I’ve seen other products becoming overly familiar, in an attempt to keep customers paying, if I had to guess.
Honestly, it’s a hundred times worse than what I can manage on my 5090 GPU, and I stopped bothering with local setups ages ago because the sheer number of bugs was just too much to handle.
Unfortunately, nothing is even from the same planet right now compared to the deep research of a GPT Pro subscription. It genuinely performs PhD-level research.
You can’t block an Ad from a block of text if it is streamed from a server tho.
No way to differentiate if a paragraph is skewed towards a company/product.
Nothing blockable by a typical blocker here because those are about external scripts/blocks on the page.
Ads won’t be part of the chatbot’s response according to the OP, so uBlock will certainly be effective at identifying and removing them. However, I agree that there is a real possibility that in the future ads will be more insidiously served as part of the response itself (e.g., when asking for product reviews), in which case ad blockers will become useless.
