Better prices. One key. No subscription.
No credit card required · $5 on signup — Google, GitHub, or email
# pip install openai from openai import OpenAI client = OpenAI( api_key="bai-xxxxxxxx", base_url="https://api.boundlessapi.com/v1" # ← change this line ) response = client.chat.completions.create( model="claude-sonnet-4-6", messages=[{"role": "user", "content": "Hello!"}], ) print(response.choices[0].message.content)
We do not store your prompts or completions by default — only billing metadata (model, token counts, latency). Opt-in request logging is off unless you enable it.
Read data policy →Same flow as OpenRouter — sign up, add credits, paste your key. Works with any OpenAI SDK.
Create your account in seconds. No sales call, no approval wait.
Pay-as-you-go in USD from $9. Sign up with Google, GitHub, or email — $5.00 added to your balance. No subscription required.
Drop it into the OpenAI SDK. Change one line: the base URL.
OpenRouter-compatible unified API — same DX, better prices, one key for all models.
Claude, GPT, Gemini, DeepSeek & more — one key, one endpoint, one bill.
Published $/1M rates vs official and OpenRouter — no platform fees on top.
Route across providers when one goes down. Your app keeps running.
Buy credits, use credits. No monthly plans, no lock-in.
Credits never expire. Builder +5% bonus credits; Startup +25%. No OpenRouter-style platform fee.
Typical gpt-5.5 chat (~3K in + 500 out): ~$0.0075 on Boundless vs ~$0.015 official (50% off). Full pricing →
Boundless, official, and OpenRouter rates side by side. OpenRouter = official + 5%. GPT & Claude −50% vs official · others −25%.
| Model | Boundless / 1M | Official / 1M | OpenRouter / 1M | vs Official | Status |
|---|---|---|---|---|---|
| claude-sonnet-4-6 POPULAR | $1.5 / $7.5 | $3 / $15 | $3.15 / $15.75 | −50% | Live |
| gpt-5.5 | $1.25 / $7.5 | $2.5 / $15 | $2.62 / $15.75 | −50% | Live |
| claude-opus-4-8 | $2.5 / $12.5 | $5 / $25 | $5.25 / $26.25 | −50% | Live |
| claude-opus-4-7 | $2.5 / $12.5 | $5 / $25 | $5.25 / $26.25 | −50% | Live |
| gpt-5.4 | $1.25 / $7.5 | $2.5 / $15 | $2.62 / $15.75 | −50% | Live |
| claude-haiku-4-5-20251001 | $0.4 / $2 | $0.8 / $4 | $0.84 / $4.2 | −50% | Live |
6 live models now · 42 in catalog · Browse full catalog →
We win on price and simplicity for most indie use cases. Some niche models may be cheaper elsewhere — we tell you upfront.
Fully OpenAI-compatible. Your existing SDK code works as-is.
# pip install openai from openai import OpenAI client = OpenAI( api_key=os.environ["BOUNDLESS_API_KEY"], base_url="https://api.boundlessapi.com/v1", ) stream = client.chat.completions.create( model="claude-sonnet-4-6", messages=[{"role": "user", "content": "Hello, Boundless."}], stream=True, ) for chunk in stream: print(chunk.choices[0].delta.content or "", end="")
No fake logos. No unverified claims. Just transparent policies you can read.
Every model shows Boundless, official, and OpenRouter $/1M rates. No hidden platform fees.
View pricing table →Real-time uptime and per-model availability. Incident history posted publicly.
System status →Zero prompt retention by default. TLS in transit, encrypted key storage.
Security docs →Prompts & completions are not stored by default — only billing metadata (model, token counts). Opt-in logging is off unless you enable it.
Data policy →US users: US-region inference by default. EU & Southeast Asia: nearest available endpoint. US API traffic is not routed through mainland China.
Regional routing →Same OpenAI-compatible API and self-serve flow. Boundless offers 42 curated models — GPT & Claude 50% off official, others 25% off — with side-by-side OpenRouter comparison and no platform fee on credits. Plus $5 on signup — Google, GitHub, or email. Migration takes one line of code.
Point your base_url to https://api.boundlessapi.com/v1 and swap your API key. If you use the OpenAI Python or TypeScript SDK, that's literally one line.
API content: By default we do not store your prompts or completions — they are processed in memory and forwarded to the model provider. We never use your content for training.
Metadata: We do retain billing metadata (timestamp, model ID, token counts, latency) for up to 13 months — no prompt text unless you opt in to request logging in Dashboard → Settings.
Providers: Upstream model providers have their own policies; see our Model Provider Terms Index.
Buy credits via Stripe (USD), pay per token as you go. No subscription, no OpenRouter-style platform fee. Each model lists Boundless vs official vs OpenRouter rates. Builder +5% bonus credits; Startup +25%.
6 models live today (Claude & GPT) with 42 in the full catalog. Use exact model IDs — see Quickstart or Models. Names with spaces or wrong casing will fail.
United States: Inference is routed to US-region endpoints by default (billing entity: Boundless Intelligence Labs, Inc., Delaware).
EU & Southeast Asia: Traffic is routed to the nearest available region for low latency. Dedicated EU region locking is available on Enterprise plans; self-serve EU routing is on our roadmap.
China: We do not route US customer API traffic through mainland China for inference or storage.
Details: Data residency & routing.
Code EARLY50 stacks an extra 50% discount on top of our standard rates (75% off official for GPT/Claude, 25% off for other models). It is valid through December 31, 2026 — enter it at checkout in the dashboard. Purchased credits never expire while your account is active. Need a longer window for your team? Email support@boundlessapi.com.
Join our Discord community — ask in #support for quick help. You can also email support@boundlessapi.com. See our support policy for response SLAs.
$5.00 on signup · No credit card · First API call in under 5 minutes
Need a custom SLA or volume pricing? Talk to us — enterprise is a secondary path, not the default.