DeepSeek V3 is easily the best cost / performance non-reasoning model on the market but accessing it via API is not straightforward -- at least for people who want their model hosted in the US. For the layperson OpenAI / Anthropic are much more accessible.
Some of OpenRouter's [0][1] providers seem bunk; their version is corrupted. Occasionally just get mangled outputs. Fireworks [2] is my best experience so far: their serverless API, haven't messed with custom deployments. Fast, performant, better than other models in its size category and less expensive.
You can save one life for a lot cheaper than the countless billions poured into OpenAI. This might be the least efficient life-saving intervention in history, and that's before calculating for harm caused.
It costs UNICEF about $4 for a single malaria vaccine[1].
"Globally in 2023, there were an estimated 263 million malaria cases and 597 000 malaria deaths in 83 countries." [2]
" In the 20th century alone, malaria claimed between 150 million and 300 million lives, accounting for 2 to 5 percent of all deaths" [3]
There's a widely claimed factoid that Malaria has killed half of all people ever born, and while that might be true, there's plenty of debate you can find about how accurate that claim is. In any case, Malaria is one of, if not the, most deadly infectious disease in human history.
We could have wide-spread elimination of malaria transmission now. We could eradicate malaria with further developments. [4]
> IMO, saving one life justifies every cent invested into OpenAI.
Common reporting says openAI spent $5b last year. That's 1.25b vaccines. The population of Africa is 1.5b, and Africa has the most malaria cases of any continent. So you could make a really, really big impact in that 263m cases/597k deaths per year just by investing in something that already exists.
>We could have wide-spread elimination of malaria transmission now.
The vaccine doesn't work that well. As your article [4] says
>The highest impact will be achieved, however, when the vaccines are introduced alongside a mix of other WHO-recommended malaria interventions such as bed nets and chemoprophylaxis.
> The vaccine doesn't work that well. As your article [4] says
It does not say that.
And in any case, as we should all very well know by now, vaccines don't have to be 100% effective at disease prevention to be effective at disease spread reduction, especially at scale.
But, lets grant the point: Fine. Use some of that $5b annual spend to fund other preventatives.
Oops bad punctuation - I should have had a colon. The text after the > is cut and paste from the article. From wikipedia "fourth dose extends the protection for another 1–2 years. The vaccine reduces hospital admissions from severe malaria by around 30%" So it's a helps a bit situation rather than just vaccinate everyone and malaria is over.
I think an argument could be reasonably made that the app layer is the only moat. It’s more likely Anthropic eventually has to acquire Cursor to cement a position here than they out-compete it. Where, why, what brand and what product customers swipe their credit cards for matters — a lot.
(1) That's a big if. It requires building a team specialized in delivering what Cursor has already delivered which is no small task. There are probably only a handful of engineers on the planet that have or can be incentivized to develop the product intuition the Cursor founders have developed in the market already. And even then; I'm an aspiring engineer / PM at Anthropic. Why would I choose to spend all of my creative energy copying what somebody else is doing for the same pay I'd get working on something greenfield, or more interesting to me, or more likely to get me a promotion?
(2) It's not clear to me that users (or developers) actually behave this way in practice. Engineering is a bit of a cargo cult. Cursor got popular because it was good but it also got popular because it got popular.
In my opinion you're vastly overestimating how much of a moat Cursor has. In broad strokes, in builds an index of your repo for easier referencing and then adds some handy UI hooks so you can talk to the model, there really isn't that much more going on. Yes, the autocomplete is nice at times, but it's at best like pair programming with a new hire. Every big player in the AI space could replicate what they've done, it's only a matter of whether they consider it worth the investment or not given how fast the whole field is moving.
If Zed gets its agentice editing mode in I’m moving away from Cursor again. I’m only with them because they currently have the best experience there. Their moat is zero, and I’d much rather use purely API models than a Cursor subscription.
> It requires building a team specialized in delivering what Cursor has already delivered which is no small task.
There are several AIDEs out there, and based on working with Cursor, VS Code, and Windsurf there doesn't seem to be much of a difference (although I like Windsurf best). What moat does Cursor have?
Just chiming in to say that AIDEs (Artificial Intelligence Development Environments, I suppose) is such a good term for these new tools imo.
It's one thing to retrofit LLMs into existing tools but I'm more curious how this new space will develop as time goes on. Already stuff like the Warp terminal is pretty useful in day to day use.
Who knows, maybe this time next year we'll see more people programming by voice input instead of typing. Something akin to Talon Voice supercharged by a local LLM hopefully.
And Typescript simply doesn't work for me. I have tried uninstalling extensions. It is always "Initializing". I reload windows, etc. It eventually might get there, I can't tell what's going on. At the moment, AI is not worth the trade-off of no Typescript support.
My entire company of 100+ engineers is using cursor on multiple large typescript repos with zero issues. Must be some kind of local setup issue on your end, it definitely works just fine. In fact I've seen consistently more useful / less junky results from using LLMs for code with typescript than any other language, particularly when cursor's "shadow workspace" option is enabled.
They do actually have custom models for autocomplete (which requires very low latency) and applying edits from the LLM (which turns out to require another LLM step, as they can’t reliably output perfect diffs)
Something worth noting is that ChatGPT currently is the killer app -- DeepSeek's current chart-topping app notwithstanding (not clear if viral blip or long-term trend).
ChatGPT Plus gives me a limited number of o1 calls and o1 doesn't have web access, so I mostly have been using 4o in the last month and supplementing it with DeepSeek in the last week, for when I need advanced reasoning (with web search in DeepSeek as a bonus).
The killer app had better start giving better value, or I'd gladly pay the same amount of DeepSeek for unlimited access if they decided to charge.
For me ChatGPT was not that useful for work, the killer app was Cursor. It’ll be similar for other industries, it needs to be integrated directly in core business apps.
I just wanted to comment to say that the design of this blog / personal website is incredible. A very minimalistic, nostalgic, and yet modern take on personal websites from the Geocities era.
Similar to blog post, instead of at the extension layer I built a PostgreSQL ORM for Node.js based on ActiveRecord + Django's ORM that includes the concept of vector fields [0][1] that lets you write code like this:
// Stores the `title` and `content` fields together as a vector
// in the `content_embedding` vector field
BlogPost.vectorizes(
'content_embedding',
(title, content) => `Title: ${title}\n\nBody: ${content}`
);
// Find the top 10 blog posts matching "blog posts about dogs"
// Automatically converts query to a vector
let searchBlogPosts = await BlogPost.query()
.search('content_embedding', 'blog posts about dogs')
.limit(10)
.select();
I find it tremendously useful; you can query the underlying data or the embedding content, and you can define how the fields in the model get stored as embeddings in the first place.
I haven’t looked at the code but it’s fairly likely the author considered this? eg the new timeout is set based on the delta of Date.now() instead of just subtracting the time from the previous timeout.
To be fair, this is what I expect of any delay function. If it needs to be precise to the millisecond, especially when scheduled hours or days ahead, I'd default to doing a sleep until shortly before (ballpark: 98% of the full time span) and then a smaller sleep for the remaining time, or even a busy wait for the last bit if it needs to be sub-millisecond accurate
I've had too many sleep functions not work as they should to still rely on this, especially on mobile devices and webpages where background power consumption is a concern. It doesn't excuse new bad implementations but it's also not exactly surprising
But it appears that it is consistent with setTimeout’s behavior and therefore likely correct in the context it will be used.
At least if your definition of “correct” is “does the thing most similar to the thing I’m extending/replicating”. In fact you might believe it’s a bug to do otherwise, and JS (I’m no expert) doesn’t give a way to run off the event loop anyway (in all implementations). Although I’d be amused to see someone running even a 90 day timer in the browser. :)
I’ve think a very precise timeout would want a different name, to distinguish it from setTimeout’s behavior.
As I understand it, the precision of such timers has been limited a bit in browsers to mitigate some Spectre attacks (and maybe others), but I imagine it would still be fine for this purpose.
Off the top of my head, a cron scheduler for a server that reads from a database and sets a timeout upon boot. Every time the server is reboot the timeouts are reinitialized (fail safe in case of downtime). If upon boot there’s a timeout > 25 days it’ll get executed immediately which is not the behavior you want.
Not speaking for OpenAI here, only myself — but this is not an official SDK — only a reference implementation. The included relay is only intended as an example. The issues here will certainly be tackled for the production release of the API :).
I’d love to build something more full-featured here and may approach it as a side project. Feel free to ping me directly if you have ideas. @keithwhor on GitHub / X dot com.
Some of OpenRouter's [0][1] providers seem bunk; their version is corrupted. Occasionally just get mangled outputs. Fireworks [2] is my best experience so far: their serverless API, haven't messed with custom deployments. Fast, performant, better than other models in its size category and less expensive.
[0] https://openrouter.ai/deepseek/deepseek-chat:free
[1] https://openrouter.ai/deepseek/deepseek-chat
[2] https://fireworks.ai/models/fireworks/deepseek-v3
reply