Skip to content
Live

ozmio-1-fast is 2.1× faster

POST /v2/chat/completions — 200 OK · 142 ms

One API. Every model. Zero cold starts.

OpenAI-compatible inference for 60+ frontier and open models, served from 14 edge regions with automatic fallbacks. Change one line, pay per token.

playground · api.ozmio.dev

Model

curl https://api.ozmio.dev/v2/chat/completions \
  -H "Authorization: Bearer $OZMIO_KEY" \
  -d '{
    "model": "ozmio-1-fast",
    "stream": true,
    "messages": [{ "role": "user",
      "content": "Name 3 uses for edge inference" }]
  }'

Response · stream

waiting

TTFT
—
Tokens / s
—
Tokens
0
Cost
$0.00000
Median time to first token
0ms
Models, one schema
0
Edge regions
0
Tokens served yesterday
0T

01 Models

The catalog, priced per token

Sort by price, speed or context. Every model speaks the same request format.

Model catalog. Use the column buttons to sort. 8 models shown.
Type
ozmio-embed-3Embeddings8K$0.02—9,000 t/s
ozmio-1-fastNewText128K$0.10$0.401,840 t/s
open-8b-instructText32K$0.12$0.122,100 t/s
ozmio-coder-2Code256K$0.30$1.20920 t/s
open-70b-instructText128K$0.59$0.79640 t/s
ozmio-vision-1Vision64K$0.80$2.40410 t/s
ozmio-1-largeText200K$2.00$8.00310 t/s
open-405b-reasoningCode128K$3.00$3.00120 t/s

Prices in USD. Cached input tokens are billed at 25%. Batch requests are 50% off.

02 Benchmarks

Fast where it counts

Same 70B model, same prompt, measured from 6 continents every 10 minutes. Methodology and raw data are public.

See live status

Time to first token

p50 · ms · lower is better
  • Ozmio edge142
  • Provider A388
  • Provider B512
  • Self-hosted GPU706

2.7×

faster TTFT

640

tokens / s

−58%

cost vs. A

03 Platform

Production plumbing, included

Drop-in compatible

Keep your SDK. Swap the base URL and every endpoint just works.

- baseURL: "https://api.other.com/v1"
+ baseURL: "https://api.ozmio.dev/v2"

Routing & fallbacks

Route by cost or latency and fail over across models and regions automatically.

model: ["ozmio-1-fast", "open-70b"]
route: "lowest-latency"

Prompt caching

Long system prompts are cached at the edge. Hits cost 75% less and return faster.

x-ozmio-cache: HIT  saved: 11,204 tokens

Structured outputs

Pass a JSON schema; get valid JSON back. Guaranteed by constrained decoding.

response_format: { type: "json_schema" }

Batch API

Queue millions of requests and get results within 24 hours at half the price.

POST /v2/batches  →  −50%

Zero data retention

Prompts and completions are never stored or used for training. SOC 2 Type II.

retention: "none"  region: "eu-only"

04 Status

99.99% uptime, in public

All systems operational

  • API99.99% uptime
  • Streaming99.98% uptime
  • Embeddings100% uptime
  • Dashboard99.95% uptime

Regions

Live p50 latency from your nearest PoP

  • fra-1 · Frankfurt38 ms
  • iad-1 · Virginia41 ms
  • sin-1 · Singapore44 ms
  • bom-1 · Mumbai52 ms
  • gru-1 · São Paulo96 ms
  • syd-1 · Sydney47 ms

05 Pricing

Per token. Nothing else.

No seats, no minimums. Commit for volume discounts when you’re ready.

50,000
1,200
300

Estimated monthly cost

$0

0 tokens / month

  • Pay as you golist price
  • Scale · $5k / mo commit−15%
  • Enterprise · reserved GPUscustom
Start with $10 free credit

Shipping on Ozmio in production

quorbit

“We cut p95 latency from 2.4 s to 610 ms by switching one base URL. Our autocomplete finally feels instant.”

Daniel KovačStaff Engineer

Lumenfield

“Fallbacks saved us during two provider outages last quarter. Customers never noticed a thing.”

Amara OkaforHead of AI

Halvard

“Batch plus caching took our nightly enrichment job from $9k to $2.1k a month, same model.”

Maya RahmanVP Data

Developer FAQ

Is it really OpenAI-compatible?
Chat, completions, embeddings, tools, JSON mode and streaming all follow the same schema. Most apps switch by changing the base URL.
What are the rate limits?
New accounts start at 600 requests and 2M tokens per minute. Limits rise automatically with usage, or instantly on Scale.
Can I deploy a fine-tuned model?
Yes. Upload LoRA adapters or full weights and serve them on shared or dedicated GPUs with the same API.
Where is my data processed?
In the region you pick. EU-only and US-only routing are one header away, and nothing is stored after the response.

Your first token in 60 seconds

$10 of free credit. No card required to start.

Create API key

terminal

$ npm i openai
$ export OZMIO_KEY=oz_live_••••••••
$ npx ozmio chat "hello"