Every AI model,
behind one API.
GPT, Claude, Gemini, Veo, Kling, Seedance, Flux, Suno and more — one key, one invoice, one request format. The same models you already use, about 30% below official provider pricing.
- No provider contracts
- Pay per call, no minimums
- OpenAI-compatible SDKs
One key, every provider
+13 more99.9%
Uptime
24.6s
Median response
24/7
Engineer support
SOC 2
Data security
API Market
The same models. Better prices.
Every endpoint is production-ready the moment you sign up — no waitlists, no per-provider negotiation, no separate invoices.
GPT-5.2
PopularOpenAI
OpenAI's frontier model for general reasoning and tool use
GPT-5.2 Mini
Best priceOpenAI
Most of the capability at a fraction of the price
o5 Reasoning
OpenAI
Extended deliberation for maths, proofs and planning
Claude Opus 4.6
PopularAnthropic
Long-context reasoning and agentic coding
Claude Sonnet 4.6
Anthropic
The balanced Claude — fast enough for interactive apps
Claude Haiku 4.5
FastestAnthropic
Sub-second Claude for high-throughput pipelines
Gemini 3 Pro
Multimodal understanding over very long inputs
Gemini 3 Flash
Best priceFastestHigh throughput with a million-token window
Gemini Omni Flash
NewRealtime multimodal in and out
Integration
Three minutes from key to first render.
One authentication scheme, one error format and one billing balance across every model — whether you are calling a video model, an image model or a chat completion.
- 1
Create a key
Sign up and mint an API key from the dashboard. No sales call.
- 2
Point at our base URL
Swap api.openai.com for api.viddly.io — your SDK keeps working.
- 3
Ship
One balance covers every model. Usage and logs land in the dashboard live.
curl -X POST https://api.viddly.io/v1/chat/completions \
-H "Authorization: Bearer $VIDDLY_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-5-2",
"messages": [
{
"role": "user",
"content": "Explain retrieval augmented generation in two sentences."
}
],
"max_tokens": 512
}'Why viddly.io
Built for teams shipping AI in production.
~30% below official pricing
We buy capacity in bulk and keep a thin margin. Same weights, same output, smaller invoice.
One key, every provider
Stop juggling six dashboards and six contracts. Rotate, scope and revoke keys from one place.
Automatic failover
If an upstream region degrades, requests reroute mid-flight. You see latency, not errors.
Async jobs with webhooks
Long video renders return a job id immediately and POST back on completion. No polling loops.
Usage you can actually audit
Per-key, per-model spend with request logs and replayable payloads, exportable as CSV.
New models within days
When a provider ships, we wrap it. Same request shape, so adopting it is a one-line change.
Pricing
Prepay credits. Spend them anywhere.
One credit is $0.005 of API usage on any model in the market. The rate is identical in every pack — no seats, no monthly minimum, no expiry.
Starter
Try any model.
$5one-off
1,000 credits
- All models, no gating
- Unlimited API keys
- Full request logs + CSV export
- Webhooks and async jobs
- Same rate limits and priority
- Credits never expire
Builder
Side projects.
$10one-off
2,000 credits
- All models, no gating
- Unlimited API keys
- Full request logs + CSV export
- Webhooks and async jobs
- Same rate limits and priority
- Credits never expire
Spending more than $2,000 a month? Talk to us about committed-use rates →
Questions developers actually ask.
Is this the same model, or a distilled copy?+
The same weights, served by the provider. We proxy the request, normalise the request and response shape, and bill you in credits. Output is identical to calling the provider directly.
How can the prices be lower?+
We commit to volume across every provider we resell and keep a thin margin on top — typically under 20%. You get bulk pricing without the bulk commitment.
Do credits expire?+
No. Prepaid credits stay on your balance until you spend them, and unused credits are refundable within 30 days of purchase.
What happens when a provider goes down?+
Requests fail over to a healthy region automatically. If a model is unavailable everywhere, the call returns a typed 503 and no credits are deducted.
Can I keep using the OpenAI or Anthropic SDK?+
For text models, yes — change the base URL and the key and everything else works, including streaming and tool use. Video, image and audio use our own REST resources.
Where is my data stored?+
Prompts and outputs are retained only as long as your log window (30 days to 1 year by plan), and are never used for training. Zero-retention mode is available on request.