Anthropic1M
Claude Opus 4.8
NEW$5 / $25−70%
$1.5 / $7.5
input / output · USD per 1M tokens
cache hit $0.15·save 70%
Limited time · save up to 95%
One API key for
Claude, GPT, and Gemini.
Built first for coding agents. Use the SDKs your tools already speak — Claude Code, Cursor, Codex, Cline, aider, and OpenAI-compatible clients.
OpenAI-compatible·No card required·Pay per token
curl https://api.omniakey.com/v1/chat/completions \ -H "Authorization: Bearer your-omniakey-api-key" \ -H "Content-Type: application/json" \ -d '{ "model": "gpt-5.5", "messages": [{"role": "user", "content": "Hello"}] }'
Models · Three flagships
Three flagships, ready on day one.
Anthropic, OpenAI, Google — one top pick per vendor. Each card shows what you actually pay per 1M input / output tokens.
OpenAI1M
GPT-5.5
$5 / $30−95%
$0.25 / $1.5
input / output · USD per 1M tokens
cache hit $0.025·save 95%
Google1M
Gemini 3.1 Pro Preview
$2 / $12−20%
$1.6 / $9.6
input / output · USD per 1M tokens
cache hit $0.16·save 20%
95%
max savings
95%
GPT off
70%
Claude off
20%
Gemini off
Why OmniaKey
An aggregator that respects
both your time and your bill.
01
Same models, up to 95% less.
Provider-specific launch pricing: save 95% on GPT, 70% on Claude, and 20% on Gemini. No surprise markups, no tiered "credits," no special tokens. The model you ask for is the model you get.
02
One key, every SDK.
OpenAI-compatible and Anthropic-compatible endpoints. Drop OmniaKey into Cursor, Claude Code, Codex, Aider, your shell scripts — they don't know the difference.
03
Built for coding.
Quotas, routing, and rate limits tuned for the way Claude Code and Codex actually hammer a model. Long contexts and tool loops are first-class, not edge cases.
Pricing · Per token · limited-time launch promo
No subscriptions. No tiers.
Each listed model priced line by line.
See the exact input, output, and cache-hit price before you call. Add credit, use models, pay per token.
ModelOfficial $/1MOmniaKey $/1MSaveContext · capability
Claude Opus 4.8claude-opus-4-8 · Anthropic
$5 / $25cache $0.5
$1.5 / $7.5cache $0.15
−70%1M contextvision · tool use
Claude Sonnet 4.6claude-sonnet-4-6 · Anthropic
$3 / $15cache $0.3
$0.9 / $4.5cache $0.09
−70%1M contextvision · tool use
GPT-5.5gpt-5.5 · OpenAI
$5 / $30cache $0.5
$0.25 / $1.5cache $0.025
−95%1M contextvision · tool use
GPT-5.4gpt-5.4 · OpenAI
$2.5 / $15cache $0.25
$0.125 / $0.75cache $0.0125
−95%1M contextvision · tool use
Gemini 3.1 Pro Previewgemini-3.1-pro-preview · Google
$2 / $12cache $0.2
$1.6 / $9.6cache $0.16
−20%1M contextvision · tool use
Gemini 3 Flash Previewgemini-3-flash-preview · Google
$0.5 / $3cache $0.05
$0.4 / $2.4cache $0.04
−20%1M contextvision · tool use · fast
All prices in USD per 1M tokens, input / output. Mirrors the provider's published rate; updates within 24h of any change.
Honest from day one
We're new. So instead of made-up traction,
here's what we can actually back up.
No fabricated user counts, no stock-photo logos. Just the three things every customer wants to know before they paste their first sk-... in a config.
3
native protocols
Anthropic · OpenAI · Gemini — tool_use, streaming, long context, all supported.
8+
coding tools
Plugs straight into the coding agents you already use.
95%
max savings
GPT, Claude, and Gemini priced separately
Drops into the tools you already use —
- Claude Code
- Codex CLI
- Gemini CLI
- Cursor
- Cline
- OpenCode
+ Aider · Continue · Zed · anything OpenAI-compatible
Questions
The ones we actually get asked.
Why up to 95% cheaper than the official price?
Launch-period promo with provider-specific discounts: save 95% on GPT, 70% on Claude, and 20% on Gemini. No tier games, no minimum spend, no annual contract.
What about after the launch promo?
Launch-period promo — and yes, that means it's not forever. We'll email you ahead of any change and keep this page in sync.
Is the model the same model?
Yes. The provider, model, and weights you ask for are what runs. No quantization, no distillation, no substitution. The bill changes; the model doesn't.
What if an upstream provider goes down?
We connect directly to Anthropic, OpenAI, and Google. If a provider is down, that provider stays down on our end too — we don't silently route to a different model to mask outages, since that would tear up the same-model promise above. The dashboard shows live status per provider; switch to whichever's still working. Our dual-region setup covers our own infrastructure failures, not theirs.
Does it work with Claude Code, Codex, and Gemini?
Yes — all three are the first scenarios we built for. There's a copy-paste env block earlier on this page; the base URL suffix differs by provider (OpenAI
/v1, Gemini /v1beta, Anthropic uses the bare URL) and the block highlights the difference. Cursor, Aider, Continue, and anything OpenAI/Anthropic/Gemini-compatible work the same way.What about latency? Are you adding a hop?
Yes, there's one routing hop. Same-region deployment puts it around 30-90ms — relative to TTFT of 200ms+ and a multi-second full response, you won't notice it.
Do you log my prompts?
No prompt or response bodies stored by default. We only keep metadata for billing and the usage dashboard — timestamps, token counts, model, latency. Email us if you need debugging logs.
Where are you based, and who's behind this?
A small team of developers who got tired of juggling three dashboards and writing the same retry loop. Infrastructure is hosted across US-East and AP-Tokyo for redundancy. Reach us at
[email protected] — replies come from real humans.How do I pay?
Credit and debit cards (handled by Stripe); crypto is on the way. Top up any amount, spend it down, no expiry. Refunds available on unused balance within 30 days.
One key.
Claude, GPT, and Gemini.
Save up to 95%.
Sign up in 30 seconds and prove it on your own code.