When I'm building apps that use LLMs I'm often looking up prices from the providers' websites, sometimes copying and pasting them into prompts ('calculate and display the costs for this model using this screenshot'). I'd rather tell Claude Code 'use toktab.com to get the current prices for gemini-3-flash-preview'. And now I can! This site is built every night from the open source pricing data from LiteLLM - https://github.com/BerriAI/litellm - which was the most comprehensive data I could find. I think the main weakness is lack of fuzzy searching, which is hard to achieve with a statically generated site, but I'm very open to suggestions.
> I'm unsure whether SQLite's extension interface is flexible enough to support this.
I think it is, if I've understood your requirements correctly. e.g. from the datasette-faiss docs:
with related as (
select value from json_each(
faiss_search(
'intranet',
'embeddings',
(select embedding from embeddings where id = :id),
5
)
)
)
select id, title from articles, related
where id = value
Can you say more? Usually projects that gravitate to SQLlite are not those that require massive scale and a FAISS index of a few GB covers a lot of documents.
My dataset is going to be around 10M documents. With OpenAI embeddings, that will be around 62GB. AFAIK SQLite should be able to handle that size, but I haven't tried.
This is not going to be my primary DB. I would update this maybe once a day and the update doesn't have to be super fast.
> It takes a while to configure a persistent database, redis, and storage on Heroku
Really? Have you actually tried this? In my experience Heroku couldn't make it any easier. You can provision a database in 60 seconds, from a Deploy to Heroku button in your Github README, or in the lovely dashboard, or on the CLI, or with two lines in heroku.yml.
We're a heavy Gitlab user and I'm a big fan of Gitlab's typically transparent communications style, but this article reflects really badly on you. I think you should take it down.
Wagtail - https://wagtail.io - is the most popular Python CMS. It runs sites for NASA, Google, Mozilla, NHS.uk. It's open source and under very active development.
I run Torchbox, a small, private tech agency in the UK. We were commissioned by NHS Digital (the state sector organisation behind these tools) to support two related projects: moving NHS.uk to https://github.com/wagtail/wagtail and setting up https://beta.nhs.uk/service-manual/ on Wagtail.
NHS Digital's procurement processes are tight and focused. They have an expert in-house team of developers and delivery managers, and they only used us on where we could accelerate the project.
As a business owner, I would have liked a bigger contract. As a tax-payer and frequent user of the NHS, I'm very happy with their efficiency.
The impression I get is that the NHS is pretty efficient budget wise but with a budget so large it's easy to make the inefficiencies sound bad "NHS wastes millions on Foo" sounds much worse than "NHS wastes 0.001% of it's budget on Foo".
For people who don't know the NHS budget is approx £130bn (~$165bn US), they are a huge organisation that provides healthcare to 66 million people.
"A child prodigy in chess, Hassabis reached master standard at the age of 13"
"graduating in 1997 with a Double First[13] from the University of Cambridge"
"obtain[ed] his PhD in cognitive neuroscience from University College London"
"[his] theoretical account of the episodic memory system [...] was listed in the top 10 scientific breakthroughs of the year in any field by the journal Science"
