Intelligence.In motion.

The AI gateway for your next idea.
Connect Claude Code, Codex, and your API clients. One endpoint. Every request accounted for.

Claim free credits Read the quickstart Telegram verification required
PHANTOM / SIGNAL ENGINEMODEL ROUTEREQUEST IN
POST /v1/messagesclaude-sonnet-4-6
Interactive routing illustration
Explore the gateway

One gateway for the tools you already use

Compatible clients · one API key · one usage view

Your workflow.
A new connection.

Point your client to Phantom, choose a supported model, and keep your existing workflow.

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.zokasta.com",
    "ANTHROPIC_API_KEY": "sk_live_your_key"
  },
  "model": "sonnet"
}
Example configurationKeep your key private

A request has a clear path

Your clientOne Phantom key
Model routeClaude or GPT
Usage recordTokens, status, cost

Compatible calls are routed and recorded as usage details. Prompts and responses are not included in request records.

More clarity.
Every token.

Rolling windows show the remaining amount and limit together, so you can decide what to run next.

Explore usage tracking
Usage windowsExample account
5-hour window$4.53 left

$7.00 limit · refreshes continuously

7-day window$43.70 left

$49.00 limit · refreshes continuously

Example request record200 · completed
Modelclaude-sonnet-4-6
Input1,248
Output352
Cache read768
Cache write128
Latency 1.24 s$0.01

Choose your
intelligence.

Claude for your coding agent. GPT for your next workflow. Compare the configured models and token rates, then connect through one Phantom endpoint.

12 available models · USD per million tokens
Anthropic · ClaudeConfigured

Claude Sonnet 4.6

Coding and everyday agent work

Input / 1M$3
Output / 1M$15
Cache read / 1M$0.3
Messages · Claude Code
Anthropic · ClaudeConfigured

Claude Opus 4.6 Thinking

Extended reasoning for complex tasks

Input / 1M$5
Output / 1M$25
Cache read / 1M$0.5
Messages · Claude Code
OpenAI · GPT / CodexConfigured

GPT 6.1 Sol

General coding and agent workflows

Input / 1M$3.6
Output / 1M$18
Cache read / 1M$0.18
Responses · Chat completions
OpenAI · GPT / CodexConfigured

GPT 6 Astra

Complex reasoning workflows

Input / 1M$3.6
Output / 1M$18
Cache read / 1M$0.18
Responses · Chat completions
OpenAI · GPT / CodexConfigured

GPT 6 Sol

General development workflows

Input / 1M$3.6
Output / 1M$18
Cache read / 1M$0.18
Responses · Chat completions
OpenAI · GPT / CodexConfigured

GPT 6.1 Terra

Coding and tool use

Input / 1M$3.6
Output / 1M$18
Cache read / 1M$0.18
Responses · Chat completions
OpenAI · GPT / CodexConfigured

GPT 6 Luna

Everyday assistance

Input / 1M$3.6
Output / 1M$18
Cache read / 1M$0.18
Responses · Chat completions
OpenAI · GPT / CodexConfigured

GPT 6 Chat Thinking

Conversational reasoning

Input / 1M$3.6
Output / 1M$18
Cache read / 1M$0.18
Responses · Chat completions
OpenAI · GPT / CodexConfigured

GPT 5.6 Sol

Coding and general assistance

Input / 1M$4
Output / 1M$20
Cache read / 1M$0.4
Responses · Chat completions
OpenAI · GPT / CodexConfigured

GPT 5.6 Terra

Focused coding tasks

Input / 1M$1
Output / 1M$5
Cache read / 1M$0.05
Responses · Chat completions
OpenAI · GPT / CodexConfigured

GPT 5.6 Luna

Lightweight workloads

Input / 1M$0.1
Output / 1M$0.5
Cache read / 1M$0.01
Responses · Chat completions
OpenAI · GPT / CodexConfigured

Codex Auto Review

Review code in your existing workflow

Input / 1M$3.6
Output / 1M$18
Cache read / 1M$0.18
Responses · Chat completions
Google · GeminiNot enabled

Gemini

Google models will appear here when enabled in the gateway.

Input / 1M—
Output / 1M—
Cache read / 1M—
Not enabled

Configured base rates for standard text requests. Long-context and cache-write rates can differ. Model availability follows the gateway catalog; Gemini is not currently enabled.

Built around
the way you work.

From the first request to the last token, your account brings access, visibility, and spending controls together.

Keys you control.

Create separate keys for your tools. Name, rotate, and revoke them from your account without changing how your clients work.

Manage API access

Every token has a place.

Separate fresh input, output, cache reads, and cache writes. Understand what a long agent session actually used.

Understand your usage

Know your headroom.

Check the amount left in your 5-hour and 7-day windows. Compare both limits before starting your next workload.

Find your allowance

The numbers stay.
Your words don’t.

See request status, latency, model, token counts, cache reads and writes, and calculated cost. The gateway stores those usage fields without storing the prompt or response text in its request records.

See the request log
Example request record200 · completed
Modelclaude-sonnet-4-6
Input1,248
Output352
Cache read768
Cache write128
Latency 1.24 s$0.01

Room for your
next big idea.

Start free. Choose more allowance as your workload grows.

Free

14-day trial - Telegram + channel verification

$0/ 14 days
Claim free credits
5-hour allowance
$7
7-day allowance
$49
API keys
3
  • Rolling usage windows
  • Token and cache accounting
  • Telegram verification required

Pro

For developers finding their orbit

$6/ month
Choose Pro
5-hour allowance
$10
7-day allowance
$70
API keys
5
  • Rolling usage windows
  • Token and cache accounting
  • Telegram community support

Scale

For solo devs and CLI agents

$10/ month
Choose Scale
5-hour allowance
$20
7-day allowance
$140
API keys
10
  • Rolling usage windows
  • Token and cache accounting
  • Standard logging - community support

Power

For small teams running steady agent workloads

$50/ month
Choose Power
5-hour allowance
$100
7-day allowance
$700
API keys
50
  • Rolling usage windows
  • Token and cache accounting
  • Priority support - 24-hour response time

Apex

For teams scaling agent swarms across regions

$100/ month
Choose Apex
5-hour allowance
$200
7-day allowance
$1,400
API keys
100
  • Rolling usage windows
  • Token and cache accounting
  • SSO-ready - multi-region failover priority

Titan

For production fleets and high-volume agent stacks

$200/ month
Choose Titan
5-hour allowance
$400
7-day allowance
$2,800
API keys
200
  • Rolling usage windows
  • Token and cache accounting
  • Dedicated capacity & SLAs

God

Maximum throughput for enterprise agent infrastructure

$500/ month
Choose God
5-hour allowance
$1,000
7-day allowance
$7,000
API keys
500
  • Rolling usage windows
  • Token and cache accounting
  • 24-hour architecture support

Subscription fees and model usage rates are separate. The model rates above determine how much allowance a request uses. Both rolling windows apply; unused allowance does not accumulate.

Start with free credits.

Verify with Telegram to claim free credits. Compare plans, usage windows, and payment options before you subscribe.

Account controlsAPI keysUsage windowsPaymentsReferrals

Common questions.

Do I need to change how I work?

Most supported clients only need the Phantom base URL, your API key, and a model ID. The quickstart has examples for Claude Code and OpenAI-compatible requests.

What appears in the usage log?

Request status, model, timing, token counts, cache usage, and calculated cost. Prompt and response text are not saved in the gateway request records.

How do I start with free credits?

Create an account and verify through the Phantom Telegram bot. The dashboard then shows your remaining usage windows.

Where can I see plan limits?

Each plan shows its 5-hour and weekly limits on the pricing page before you choose a payment option.

Build what’s next.

One endpoint for your model calls. A readable record of what each request used.

Claim free credits