frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

AutoKernel: Autoresearch for GPU Kernels

https://github.com/RightNow-AI/autokernel
2•frozenseven•2m ago•0 comments

Passkeys were supposed to replace passwords, but they're failing

https://www.howtogeek.com/passkeys-were-supposed-to-replace-passwords-but-theyre-failing-for-the-...
1•vdelitz•3m ago•1 comments

Zen Fascists Will Control You

https://www.ianbetteridge.com/zen-fascist-wi/
1•brunohaid•5m ago•0 comments

Compact-dict – a cache-local, linear probing hash map in Rust

https://github.com/gustawdaniel/compact-dict
1•gustawdaniel•6m ago•1 comments

Prisoner's Dilemma in World Baseball Classic for Mexico and Italy Threatens USA

1•travisjungroth•7m ago•0 comments

I'm going to build my own OpenClaw, with blackjack and bun

https://github.com/rcarmo/piclaw
1•rcarmo•8m ago•0 comments

I Reduced 5 hours of Testing my Agentic AI applcaition to 10 mins

https://github.com/onepaneai/mantis
1•ashmil•9m ago•1 comments

SpacetimeDB: A Short Technical Review

https://strn.cat/w/articles/spacetime/
1•quantumwoke•11m ago•0 comments

Show HN: Lightsecond Video Surveys

https://lightsecond.io/
2•dmonn•12m ago•0 comments

SpaceX launches 15K-pound TV satellite to orbit on its 30th mission of the year

https://www.space.com/space-exploration/launches-spacecraft/spacex-launches-15-000-pound-tv-satel...
2•Brajeshwar•12m ago•0 comments

Story of Weda Bay: How nature is being sacrificed for mining

https://www.theguardian.com/environment/ng-interactive/2026/mar/11/how-mining-threatens-the-world...
2•defrost•15m ago•0 comments

Download Xiaohongshu Video

https://downloadxiaohongshuvideo.com/
2•Evan233•16m ago•0 comments

C++26: The Oxford Variadic Comma

https://www.sandordargo.com/blog/2026/03/11/cpp26-oxford-variadic-comma
3•ingve•17m ago•0 comments

Axllm: DSPy for TypeScript

https://axllm.dev/
2•handfuloflight•19m ago•0 comments

Mathematics is undergoing the biggest change in its history

https://www.newscientist.com/article/2518526-mathematics-is-undergoing-the-biggest-change-in-its-...
4•jonbaer•19m ago•1 comments

Coding agents for production iOS: a senior engineer's setup for 2x the output

https://ignatovv.me/blog/coding-agents-for-production-ios/
3•tosh•19m ago•0 comments

Microsoft patents system for AI helpers to finish games for you

https://www.dexerto.com/gaming/microsoft-patents-system-for-ai-helpers-to-finish-games-for-you-33...
3•JeanKage•21m ago•0 comments

Millennium Challenge 2002

https://en.wikipedia.org/wiki/Millennium_Challenge_2002
1•vrganj•23m ago•0 comments

PromptVault free tool for multi agentic development

2•bohdokas•24m ago•0 comments

Ask HN: How to Plan for a Better GTM for a Digital SaaS Product B2B

1•sriramgonella•27m ago•0 comments

The 'number station' sending mystery messages to Iran

https://www.ft.com/content/86c4a4ca-ca06-4fc8-90fe-4f46357b804f
3•thm•28m ago•0 comments

White-Box Attacks on PhotoDNA Perceptual Hash Function

https://eprint.iacr.org/2026/486
1•_____k•29m ago•0 comments

Michael Faraday: Scientist and Nonconformist (1996)

http://silas.psfc.mit.edu/Faraday/
3•o4c•35m ago•0 comments

An open letter to Grammarly and other plagiarists, thieves and slop merchants

https://www.moryan.com/an-open-letter-to-grammarly-and-other-plagiarists-thieves-and-slop-merchants/
3•october8140•40m ago•0 comments

Reaching net zero by 2050 'cheaper for UK than one fossil fuel crisis'

https://www.theguardian.com/environment/2026/mar/11/reaching-net-zero-by-2050-cheaper-for-uk-than...
3•ljf•42m ago•0 comments

4n6Img Project

1•Hoxed•47m ago•0 comments

Capital Isn't Destiny

https://writing.nikunjk.com/p/capital-isnt-destiny
2•tosh•47m ago•0 comments

Show HN: CLI and TUI for Elasticsearch and OpenSearch

https://github.com/jillesvangurp/kt-search/tree/master/ktsearch-cli
2•jillesvangurp•47m ago•1 comments

FreeBSDKit: Swift Package to Write Capability-Aware FreeBSD Apps

https://christiantietze.de/posts/2026/03/freebsdkit-swift-package-write-capability-aware-freebsd-...
2•frizlab•49m ago•0 comments

At the Boundary of Self-Reference

https://changkun.substack.com/p/at-the-boundary-of-self-reference
3•changkun•50m ago•0 comments
Open in hackernews

Ask HN: How are people forecasting AI API costs for agent workflows?

4•Barathkanna•1h ago
I’ve been experimenting with agent-based features and one thing that surprised me is how hard it is to estimate API costs.

A single user action can trigger anywhere from a few to dozens of LLM calls (tool use, retries, reasoning steps), and with token-based pricing the cost can vary a lot.

How are builders here planning for this when pricing their SaaS?

Are you just padding margins, limiting usage, or building internal cost tracking? Also curious, would a service that offers predictable pricing for AI APIs (like a fixed subscription cost) actually be useful for people building agentic workflows?

Comments

clearloop•1h ago
imo switch to local models could be an option
Barathkanna•58m ago
Local models solve the marginal cost problem, but they move the complexity into infrastructure and throughput planning instead.
Lazy_Player82•57m ago
Honestly, if you're designing your agent workflows properly with hard limits on retries and tool calls, the variance shouldn't be that wild. Most of the unpredictability comes from not having those guardrails in place early on. A few weeks of real production data usually shows the average cost is more stable than you'd expect.
Barathkanna•54m ago
True, but for early stage builders it’s harder to design those guardrails upfront. A lot of the time you only discover the retry patterns and cost spikes once real users start hitting the system.
Lazy_Player82•49m ago
Fair point. And honestly, with more non-technical builders shipping agent-based products these days, that's probably where a service like this makes the most sense – for people who don't yet have the experience to know what guardrails to put in place.
Barathkanna•46m ago
Exactly. That’s actually why we started building Oxlo.ai. Early stage builders usually just want to experiment without worrying too much about token cost spikes.
sriramgonella•25m ago
local models are better in controlling costs rather commercial models are very high and no control on this cost..how ever again local models training setup to be archietected very well to train this continoulsly