A unified interface
for LLMs
ONE API · 143 MODELS · 0% MARKUP
QUALIFY AND GET $100 IN TRIAL CREDITS · NO COMMITMENT
# straitly · chat completions
from openai import OpenAI
client = OpenAI(base_url=
"https://api.straitly.ai/v1")
res = client.chat.completions
.create(
model=
"anthropic/fable-5",
messages=[{"role": "user",
"content": "hello"}])TRUSTED BY ENGINEERS USING
WarpWarpWarpUNDER THE HOOD
Optimized routing for
Point your app at one endpoint, name the model, and Straitly handles the rest.
TOKENS SERVED, AND COUNTING
5.6-solOPENAI
fable-5ANTHROPIC
3.1-proYOUR DATA, NOT OURS
Prompts pass through.
They don't stay.
Zero data retention
Requests and responses are never stored. Once your tokens are delivered, they're gone.
Never trained on
Your prompts never become training data. Not ours, not the providers' we route to.
SOC 2 Type II
Audited controls behind every request. The paperwork your security team asks for, done.
ENFORCED ON EVERY REQUEST · EVERY PROVIDER
Up to 30% off
your first $10K of spend.
CUSTOM RATES FOR YOUR TOKEN SPEND
Indie hackers
Solo devs shipping real products.
Small startups
Pre-seed through Series A, before the bill gets scary.
Engineering teams
Teams spending real money on inference every month.
143 MODELS · 24 PROVIDERS
Every frontier model.
Qualified rates.
Anthropic
4 MODELS· $ PER MTOK
- claude-fable-5: read $10.00, write $50.00, cache read $1.00, cache write $12.50 per million tokens
- claude-opus-5: read $5.00, write $25.00, cache read $0.50, cache write $6.25 per million tokens
- claude-sonnet-5: read $2.00, write $10.00, cache read $0.20, cache write $2.50 per million tokens
- claude-haiku-4-5: read $1.00, write $5.00, cache read $0.10, cache write $1.25 per million tokens
OpenAI
3 MODELS· $ PER MTOK
- gpt-5.6-sol: read $5.00, write $30.00, cache read $0.50, cache write $6.25 per million tokens
- gpt-5.6-terra: read $2.00, write $12.00, cache read $0.20, cache write $2.50 per million tokens
- gpt-5.6-luna: read $0.20, write $1.20, cache read $0.02, cache write $0.25 per million tokens
3 MODELS· $ PER MTOK
- gemini-3.1-pro: read $2.00, write $12.00, cache read $0.20, cache write — per million tokens
- gemini-3.6-flash: read $1.50, write $7.50, cache read $0.15, cache write — per million tokens
- gemini-3.5-flash-lite: read $0.30, write $2.50, cache read $0.03, cache write — per million tokens
Meta
1 MODEL· $ PER MTOK
- muse-spark-1.1: read $1.25, write $4.25, cache read $0.15, cache write — per million tokens
0% markup, 30% off your first $10K of spend, plus $100 in free trial credits.
FROM APPLY TO API KEY
You could be on program rates by tonight.
01
APPLY
Two minutes. Your usage and the models you need.
02
REVIEW
Our team checks your quota. Hours, not weeks.
03
GET YOUR KEY
Approved? Your key goes live with $100 in trial credits.
04
PICK YOUR PRICING
Trial smooth? Stay usage-based or commit. Heavy discounts either way.
LOADING...
FINAL STAGE
Stop paying Router tax.
Two minutes, five questions. Qualify and your key is live today with $100 in trial credits.
$100 FREE CREDITS · NO COMMITMENT · NO SALES CALL
BEFORE YOU ASK
Fair questions.
One API for every AI model. Instead of signing up with OpenAI, Anthropic, Kimi, Z.AI and everyone else separately, each with its own key and its own bill, you get a single key that reaches all of them. Requests route to the best provider for the job and fail over automatically when one goes down. We take no markup on tokens, and new accounts start with $100 in completely free credits plus 30% off their first $10K of spend.
STEP 01
Apply
Two minutes, five questions. Tell us what you're building and which models you need.
REVIEW TIME1-2 hoursSTEP 02
Get your key
Approved accounts go live the same day, preloaded with free trial credits. No card, no sales call.
BALANCE$100.00STEP 03
Swap your base URL
Point your existing OpenAI client at Straitly, then call any model in the catalog by name. Nothing else in your code changes.
BASE URL"https://api.straitly.ai/v1"STEP 04
Ship and scale
Track spend per key as you grow, then stay usage-based or commit volume for a lower rate.
FIRST $10K30% off
We hold committed capacity with the frontier labs and major open source providers, priced below list. Most companies spend their growth budget on ads. We spend ours on your inference bill. Qualified accounts get the margin, and word of mouth does the rest.
Indie hackers shipping real products, startups from pre-seed through Series A, and engineering teams with real monthly inference spend.
Yes. Every new user gets $100 in credits at the discounted rates. If the trial works out and you want to go further, stay usage-based or commit for even cheaper prices.
Yes. It's one line of change: swap the base URL, put in your API key, and your requests come to us. OpenAI-compatible API, every model in the catalog.
Yes. Bring your own provider keys and route them through us, or use ours and get one bill instead of four. Either way it's the same OpenAI-compatible API.
We review applications on a rolling basis. Current times are 1-2 hours.
Usage-based rates might change with market pricing. Committed rates never change. Lock a commit and that's your price, period.
We currently serve companies in the $2K to $100K a month range. For larger quota requirements we recommend going directly to the foundational providers - at that size you can probably secure an enterprise deal yourself.
Fallbacks are built in - requests reroute across providers before you see an error. We guarantee a 99% SLA and top-of-the-line time to first token.
That's what the trial credits are for. Test us with $100 of real traffic, and if we're not better, walk away. No commitment, nothing to cancel.