Probably more like GitHub Copilot. A flat monthly fee that's much higher than the actual costs.
You can build your own Copilot or ChatGPT with the API that OpenAI provides, and it would be about 1/50 of the cost, but most people prefer to just give money instead of digging through the fine details.
I think it's likely the other way around, a product like Copilot is actually much cheaper than if you were to build off of OpenAI's API. ChatGPT, the product, will be much cheaper than if you actually ran the API yourself.
Some context is that I work for a company that builds a product called Codeium, that's a free Copilot alternative. If we were to use OpenAI's API, we computed our costs would be high 100s of dollars a year per user (we've opted for something more custom / in house).
I suspect that OpenAI's primary source of income will be as an API to other apps. I just read that MS is looking to use it in its office suite. It doesn't take much insight to see that many other companies will want to do it too.
OpenAI's Chatgpt will have a hard time competing as a stand-alone app since the big tech players all have a version of AI that they can offer. Big tech can ultimately offered their AI services for free since they have other sources of income.
If they follow the Dall-E2 model they will charge about 2 cents per prompt. That may sound good but it's hard to plan as a business when your business' income is hard to predict.
My guess is that ChatGPT will be offered as a subscription with the lowest level somewhere between $15 and 30 dollars a month for individuals. The subscription price will scale for businesses of different sizes. They can't charge too much since other companies can come in and underprice them. They need to make it a no brainer expense for individuals.
Token based pricing will not be understood by common user. Per-question pricing is psychologically repelling. My hypothesis is that there will be 3 tiers of pricing. For example, FREE (limited to 10 questions per day. Logic is that an average person searches 3-4 times a day). PERSONAL (100 questions/day - $10/month, you will be billed additional $1 for every 300 questions). PREMIUM (1000 questions/day - $100/month, additional $50 for every 1000 questions). Similar to how cloud resources are priced. Most of us will happily pay $10/month. Developers will have to shell out $100/month. 3rd party apps will end up paying in thousands.
This has potential to be disrupted a commodity LLM (say freely available that you can embed in your devices).
Find it hard to believe developers would pay $100/month unless the product embeds deeply into their workflow. I do think it's an interesting idea to place some limitations and still have a free tier.
For me ChatGPT has already proved to be a time saver. A couple of examples:
1) I was dealing with 2 sets of data in an unfamiliar language. There was an overlap of the data. I needed to see the unique data for each set. ChatGPT gave me the solution for this.
2) I needed to interact with an RSS feed of a website. There wasn't an RSS feed for the data I was after. ChatGPT generated a program retrieving the data via the website's API and then generated a second program which converted the data to an RSS feed.
For the increased productivity, I can see how $100/month can be within developers' budgets considering their salaries.
This seems plausible. There are many content creators, developers and others that would happily allocate 100$ for the service, since they will be getting their money's worth.
14 comments
[ 3.5 ms ] story [ 44.1 ms ] threadThe ChatGPT is just them tuning it for now, so not interesting from a monetization perspective.
You can build your own Copilot or ChatGPT with the API that OpenAI provides, and it would be about 1/50 of the cost, but most people prefer to just give money instead of digging through the fine details.
Some context is that I work for a company that builds a product called Codeium, that's a free Copilot alternative. If we were to use OpenAI's API, we computed our costs would be high 100s of dollars a year per user (we've opted for something more custom / in house).
OpenAI's Chatgpt will have a hard time competing as a stand-alone app since the big tech players all have a version of AI that they can offer. Big tech can ultimately offered their AI services for free since they have other sources of income.
If they follow the Dall-E2 model they will charge about 2 cents per prompt. That may sound good but it's hard to plan as a business when your business' income is hard to predict.
My guess is that ChatGPT will be offered as a subscription with the lowest level somewhere between $15 and 30 dollars a month for individuals. The subscription price will scale for businesses of different sizes. They can't charge too much since other companies can come in and underprice them. They need to make it a no brainer expense for individuals.
Oh no, a potentially more annoying "Mr Clippy" on the horizon.
I don't know how much they would pay me, but clearly we've all been training the thing, and should get compensated as a result.
This has potential to be disrupted a commodity LLM (say freely available that you can embed in your devices).
1) I was dealing with 2 sets of data in an unfamiliar language. There was an overlap of the data. I needed to see the unique data for each set. ChatGPT gave me the solution for this.
2) I needed to interact with an RSS feed of a website. There wasn't an RSS feed for the data I was after. ChatGPT generated a program retrieving the data via the website's API and then generated a second program which converted the data to an RSS feed.
For the increased productivity, I can see how $100/month can be within developers' budgets considering their salaries.