Hacker News Reader

Comment by 0xbadcafebee | Hacker News Reader

Both Luna and Haiku are more expensive than DS4.1 Flash at API cost. But ignore that; nobody should be using API. The open weight subscriptions for DS4.1 Flash provide higher limits at lower prices than OpenAI or Anthropic subscriptions. Finally, Haiku actually is far less token efficient/ outputs more per task. No matter how you slice it, the open weight is cheaper.

Also consider that for things like cyber work, the frontier models give you nerfed results and poor performance. Whereas the open weight isn't nerfed, and I routinely get 200t/s with my subscription. Finally, DS4.1 Flash is natively multimodal, while Haiku isn't.

They're all perfectly fine models, you should use any one of them you want. But DS4.1 Flash can do more for less. (That said: GLM-5.3-Flash is even better and cheaper...)

View on HN

There is no subscription for 1st party deepseek models, the ones that exist are all fronting as middlemen for discounted openrouter providers that serve quantized models with worse cache retention. The only real option for deepseek has been API for a few months now.

oai and anthropic also subsidize the hell out of their subs compared to what you will find in smaller competitors, a $20 codex sub gets you like $100-150 usage/wk which goes way further than 2x opencode go (which would only net out to $120 of deepseek 4.1 usage a month, on top of being low performing quantized trash).

Cache is cache, you can't get better cache than... cached...

For $20/month subscription, Charm Hyper gives you $12.50 per day, for a total of $350 per month. Like I've mentioned in other comments, OpenCode Go performance and rates are terrible now, there are several better options.

Quantization is not trash, there's a year of evidence that shows Q4 provides ~4% degradation and Q8 provides ~1% degradation, and you don't need that severely quantized to gain benefits in inference performance.

> Cache is cache, you can't get better cache than... cached...

No other provider is getting the cache hit rates deepseek does or the retention. They use a unique setup that no other provider is bothering to replicate.

> Quantization is not trash, there's a year of evidence that shows Q4 provides ~4% degradation and Q8 provides ~1% degradation, and you don't need that severely quantized to gain benefits in inference performance.

Lol. Quanting at all pretty much neuters a model, just look at how astra performs now compared to at launch.

> Both Luna and Haiku are more expensive than DS4.1 Flash at API cost

Are they. Luna uses way less tokens for identical tasks so its a bit of an apples to oranges comparison.

Reply on news.ycombinator.com