Pay only for what you use.
No subscriptions. No tiers. Just top up your balance and go.
How it works
Sign up
Create an account and get $5 in free credits instantly. No credit card required.
Make requests
Use the API with any OpenAI-compatible SDK. Credits are deducted per token at provider rates.
Top up anytime
Running low? Add more credits instantly from your dashboard.
Choose your billing method
Managed Credits
We handle provider billing
- No platform fee — credits are sold at face value
- Zero markup on token prices
- Access to all models instantly
- Smart routing & prompt optimization
0%
platform fee — credits at face value
Bring Your Own Keys
Use your existing API keys
- No routing fee on BYOK requests
- No monthly request cap
- Pay providers directly
- Smart routing & prompt optimization
$0
routing fee on your own keys
You can use both methods together. ANT will prefer your own keys when available.
Why ANT saves you money
You pay provider rates, and routing keeps you off the expensive models.
Smart routing: 30-50% savings
Our classifier detects that 60% of requests don't need GPT-4 or Opus. Automatic downrouting to faster, cheaper models saves you money on every request.
Prompt optimization: 30-50% fewer tokens
ANT compresses and rewrites your prompts before they hit the LLM. Fewer tokens in = fewer tokens billed.
Future-proof your code
Use ant:auto and let the router decide. When providers release better models, you benefit instantly.
Failover included
When Anthropic is down at 3 AM, your agents don't stop. ANT automatically switches to OpenAI or Gemini.
Real Price Comparison
What you actually pay per 1M tokens — direct vs through Antbase.
| Model | Direct Price Input / Output per 1M | Antbase Price Input / Output per 1M | You Save |
|---|---|---|---|
Claude Opus 4 Anthropic | $15.00 / $75.00 | $15.00 / $75.00 | Same token price + smart routing avoids Opus when unnecessary |
Claude Sonnet 4 Anthropic | $3.00 / $15.00 | $3.00 / $15.00 | ANT auto-routes here instead of Opus for 80% of tasks |
GPT-5 OpenAI | $10.00 / $30.00 | $10.00 / $30.00 | Same price, but ANT only uses it when needed |
GPT-4.1 OpenAI | $2.00 / $8.00 | $2.00 / $8.00 | Auto-downrouted from GPT-5 for simple tasks |
GPT-4.1 mini OpenAI | $0.40 / $1.60 | $0.40 / $1.60 | Used for 60%+ of requests via ant:auto |
Gemma 4 / Llama 4 Free providers | Free | Free | $0 — completely free via ant:free |
The real savings: intelligent routing
Antbase charges the same token prices as the providers — zero markup. The savings come from not using expensive models when you don't need them.
$0
token markup
60%
of requests don't need frontier models
30–50%
average cost reduction vs hardcoded models
Example: 1M requests/month with Claude Opus direct
Direct to Anthropic
All requests hit Opus at $15/$75 per 1M tokens
~$4,500/mo
Through Antbase (ant:auto)
20% Opus, 40% Sonnet, 30% GPT-4.1 mini, 10% free models
~$1,800/mo
That's $2,700/mo saved — at provider rates, with no platform fee on top.
How ANT Compares
| Feature | Direct to Provider | OpenRouter | Antbase |
|---|---|---|---|
| Price markup on tokens | None | None | None |
| Platform fee | None | 5.5% on credits | None |
| BYOK fee | N/A | 5% after 1M req | None |
| Intelligent routing | Basic | ||
| Prompt optimization | 30-50% token savings | ||
| Automatic failover | Provider-level | Provider + model-level | |
| Cost optimization | Auto-downroute | ||
| Virtual models | 1 (auto) | 12 modes + 2 tools |
Frequently Asked Questions
Is there a platform fee on credits?
Not today. Credits are sold at face value — $100 buys $100 of credit, and it is spent at exact provider rates with no markup on tokens. We may introduce a fee of up to 5% on credit purchases in future; it would be shown at checkout before you pay, and it would never change the token rate itself.
What is BYOK (Bring Your Own Keys)?
You can use your existing API keys from OpenRouter, Anthropic, OpenAI, etc. ANT routes through your keys — you pay providers directly. We do not currently add a routing fee on BYOK requests.
Can I use both managed credits and BYOK?
Yes. If you have both set up, ANT will prefer your own keys so the spend lands on your provider account rather than your credit balance. You can also set per-model preferences.
How does prompt optimization save money?
ANT automatically compresses and rewrites your prompts before they hit the LLM. This typically reduces input tokens by 30-50%, which directly reduces your costs.
What happens when my balance runs out?
Requests will fail with an insufficient credits error. Top up anytime from your dashboard — it's instant.
Is there a minimum purchase?
No. You can top up with as little as $10.