SLoptimise
145 posts
SLoptimise
@SLoptimise
YOUR AI startup is bleeding money and you don’t even know where. I find the leak I give unique optimisation methods for YOUR Usage. Let’s work!
Katılım Şubat 2026
27 Takip Edilen3 Takipçiler
@plumehead @dbishnoi @LilithDatura Well it depends on what LLM you're using and what tasks first. But to answer your question simply, i would say haiku for simple, Opus for complex (if you use Claude)
English
@dbishnoi @LilithDatura What model would you use to bring costs down?
English
@mgdotdev @WarrenInTheBuff What model were you using for that task though? it always almost starts with the model
English
@WarrenInTheBuff Bro I just spent $200 in API credits having Claude explain the difference between eggshell and satin
My balance sheet is now so deep in the red that I can't afford new paint supplies
English
Painting goes so much faster when you just don’t give a shit about the floors
English
@kelvongx @danshipper Have you implemented anything that will make your set up more cost-efficient yet dude?
English
@danshipper We’re about to buy 100 seats and it feels ouch enough
API usage is expensive
Not looking forward to the first bill!
English
Gabbing with my barber about Codex and he’s telling me about how he’s using Claude to run a crypto trading strategy and Gemini + nano banana to learn jiu jitsu moves
Guy in the chair next to me leans over and says his company just bought 1,300 Claude seats
2026 is wild
English
@thdxr For API usage they want some ridiculous $/mtok. Is it that good?
English
@Santosh74038967 @kr0der Yeah how are you spending that much a month? Something's wrong with your set up. What is it
English
@kr0der I have checked my token usage and the corresponding API cost on my dashboard.
I'm not going to pay ~ $5,800 a month so it is either going back to coding by hand or using open source models.
English
hypothetical scenario:
all the $20, $100, $200 AI coding subscriptions are gone, no more heavy subsidisations
what setup are you using now that you have to pay API prices?
English
100%, anyone can cut at least 40% of their costs doing these simple methods
Smoke@ShivamS1123
If your API costs are killing you, you are not chunking your data, you are not caching responses, and you are definitely not using the right model for the right task. Fix one of these first.
English
Just finished a diagnostic for a 12 person AI startup.
Found $47,000 in annual savings in 45 minutes.
How did i fix it?
3 model swaps & a prompt restructure,
DONE.
English
I’m too Facebook-brained to have an unbiased take here, but I don’t understand how more software isn’t built around recurring networks of people as core infrastructure. We barely get multiplayer right in most software experiences!
Michael Mignano@mignano
English
@TheKryptoWiz @charlieholtz This is the only right way to handle API costs and be smart with your building in AI 100%
English
@charlieholtz This is the pricing pattern builders actually notice: let the subscription handle normal work, then make the jump to API costs boring when things get spicy.
English
Here's what Anthropic pricing updates mean for Conductor users:
- You can officially use your Claude sub with Conductor
- If you're on a max subscription you get $200 in credits and then can pay at API costs
- If you use Big Terminal Mode you won't be affected
We're going to keep building the best interface for the best coding agents! Excited to show you what we've been cooking🫡
English
@madhurjain @theo Expensive? okay how much is claude making you spend per month?
English
As @theo rightly put - This is Evil!
The Claude API costs are super expensive compared to any other provider.
Time to move away all my agents to GeminiCLI or OpenCode + Kimi / GLM. Any suggestions?
ClaudeDevs@ClaudeDevs
Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage. The credit covers usage of: - Claude Agent SDK - claude -p - Claude Code GitHub Actions - Third-party apps built on the Agent SDK
English
@josevalim The policy wasn't hidden - the math was. $20/month vs real API costs at agentic scale was always unsustainable. That gap had to close. Calling it 'evading' mistakes a subsidy for a promise.
English
@SLoptimise Hey, landing page is live now! Would love your honest feedback — aryanjambholkar27-ceo.github.io/LLMeter-Landin… . Once SDK is ready you'll be first to know. 🔥
English
@sanjaybuilds_ @WorldClawAI that's actually helpful in this market
English
Quick math on AI API costs that is eating up your wallet
Running Claude Opus 4.7, GPT-5.5, and Gemini 3.1 Pro at the same time? You're paying full price on all three. Plus juggling three accounts, three API keys, three billing cycles
@WorldClawAI fixes that
WorldRouter runs 300+ models through one account at 30% below official pricing:
Opus 4.7: $3.5 / $17.5 (vs $5 / $25)
GPT-5.5: $1.75 / $10.5 (vs $2.5 / $15)
Gemini 3.1 Pro: $1.4 / $8.4 (vs $2 / $12)
Even cheaper than aggregators like B(.)AI by 35-50%
Fund one account in USD1. Every model, all in one and you won't need to platform hop anymore
As devs get smarter about API costs, services like this are going to be in serious demand
English
@FarAICoder @alxui_ux Have you tried to optimize your costs?
Well, depends on what you're building
English
English
@NathanWilbanks_ @Teknium How do you optimize the costs? That looks very costly indeed if using an API service rather than a subscription
English
I literally run 12 hermes agent instances every day in parallel to build Hermes Agent, and its now a top 100 GitHub repositories of all time. Agents do bring value and do create substantive software and work.
David Cramer@zeeg
Everyone is slowly coming to this realization, and I assure you, no one is running multitudes of agents overnight. No one that is doing anything of substance at least. There _are_ people pretending to be scientists, or fully caught up in their drug infused AI overdose, that think their slop machines are changing the world. They're not tho, and they're just wasting a bunch of money and compute to create a lot of LoC that will just get thrown away. The state of the art is still "can we even one shot a production quality patch that we wont regret later", and its rarer than you'd expect based on discourse.
English
@robinebers @nkgoutham Wow, What models are you currently using?
and for what tasks?
curious.
English
@nkgoutham luckily i only pay about $500-600 overall
the rest is free credits from labs to test their apps
but realistically i burn through $5-6k a month of subsidized usage
we’re definitely heading for a big wealth gap
English
uhm, Cursor just updated their Bugbot pricing
I know this is inevitable but… OUCH
obviously was the deal of a lifetime for the best AI reviewer, but this is going to exclude a lot of people from June 14th
we really need cheaper + good models FAST
English