Drop-in replacement · one line to migrate from OpenAI
−65% per token, same SDK
- client = openai.OpenAI(api_key="...")
+ client = openai.OpenAI(
+ api_key="thp_live_...",
+ base_url="https://api.throughputs.ai/v1", # ← only change
)
response = client.chat.completions.create(
model="your-model-id",
messages=[{"role": "user", "content": "Hello!"}],
)// PRODUCTION GRADE
Built for teams that can’t afford downtime or markups.
One routing layer, every major provider, transparent per-token pricing. Engineered for the workload you actually run — not the demo your provider's sales rep shows.
UPTIME SLA
Rolling 90-day average
60–70%
AVG SAVINGS
Vs. direct provider list price
ROUTING P95
Failover to cheapest qualified
TOKENS / SEC
Peak aggregate throughput
TO FIRST REQUEST
Signup to working call
// WHY SWITCH
Six reasons teams leave OpenAI for us
Every row below is the reason a team migrated last month. The first one — price — usually pays for the switch in week one.
PRICE
60–70% cheaper per token, day one
Bulk-negotiated wholesale rates from every major provider, passed through at per-token granularity. No markup, no monthly minimums, no commit.
−65%MEDIAN SAVINGS
MIGRATE
One line to switch from OpenAI
Change base_url. Your existing OpenAI SDK keeps working — same shape, same retries, same streaming. Zero code rewrite, zero vendor lock-in.
1LINE DIFF
KEYS
One credential, every provider
Stop juggling keys for OpenAI, Anthropic, Google, Meta, Mistral, Cohere. Integrate once, route to any model, swap with one string change.
1API KEY
RELIABILITY
Failover before you notice
If a provider wobbles, we reroute to the next cheapest qualified provider in 7ms. Your users see a response, not an error.
7MS P95
CONTROL
Per-key spend caps
Real-time token tracking, per-key budgets, team guardrails. Set a $50 cap on staging, $500 on prod, $5 on a teammate. Never get a surprise bill.
$PER-KEY CAPS
FLEXIBILITY
No lock-in, change one line to leave
Standard OpenAI-compatible API. Swap base_url back to OpenAI, Anthropic, or anywhere else in one line. No proprietary SDKs, no walled garden, no migration tax.
1LINE TO LEAVE
From signup to first request in 5 minutes
If you have a working OpenAI integration, you already have a working THROUGHPUTS integration. Change one line. That's the whole migration.
✔ Dependencies verified — nothing new to install.✔ Key authenticated · test credits applied.✔ 200 OK · 412ms · 1.2k tokens · $0.00011301
Sign up — 60 seconds
Email + password. We hand you test credits the moment you land in the dashboard.
02
Copy your API key
One credential unlocks every model we route. Copy once, paste into your existing SDK.
03
Change base_url
Point your existing OpenAI SDK at api.throughputs.ai. One line. Your code keeps working — prompts, retries, streaming all unchanged.
04
Watch your bill drop
Pay-as-you-go per-token billing at list price. Real-time spend tracking, per-key caps, no surprise invoices.
Stop paying retail markup.
Whatever you ship, you ship on tokens. Buy them at list price.
01
SaaS founders
Embed AI features without locking into one provider or paying retail markup. A/B test models, cut your bill 60–70% from day one.
02
Engineering leads
Consolidate all AI spend under one account. Per-team budgets, real-time spend tracking, switch providers without renegotiating contracts.
03
Automation builders
Power n8n, Make, or custom pipelines with the best model for each step — all from one key.
04
Researchers & evaluators
Run side-by-side benchmarks across every model without juggling 5 accounts. Same prompt, every provider, one tab.