Performance

Purple Flea API
Speed Test

We benchmark every endpoint continuously. Sub-50ms median latency on agent-native operations. Faster than Binance and dYdX for the workflows agents actually run.

12ms
Auth Median
28ms
Trade Median
45ms
Registration
99.97%
30d Uptime

Endpoint Latency

Median response times measured from Frankfurt, EU. P95 values are typically 1.8x median. All times in milliseconds. Last updated 2026-03-06.

Endpoint Path Median P95
Agent Registration POST/agents/register
45ms
81ms
Auth / Token POST/auth/token
12ms
22ms
Open Trade POST/trading/order
28ms
50ms
Wallet Balance GET/wallet/balance
18ms
33ms
Casino Play (Crash) POST/casino/crash/play
35ms
63ms
Escrow Create POST/escrow/create
42ms
76ms
Faucet Claim POST/faucet/claim
38ms
68ms
Domain Search GET/domains/search
22ms
40ms

Purple Flea vs Competitors

Latency benchmarks for equivalent operations. Purple Flea is optimised for programmatic agent access, not human web UIs.

🟣
Purple Flea
Agent-native infrastructure
Fastest
Registration45ms
Auth12ms
Trade / Order28ms
Balance Query18ms
Escrow / Settlement42ms
Casino Bet35ms
B
Binance
CEX trading API
RegistrationKYC required
Auth38ms
Trade / Order85ms
Balance Query45ms
Escrow / SettlementNot available
Casino BetNot available
Δ
dYdX
Decentralised perpetuals
Registration120ms (chain tx)
Auth52ms
Trade / Order67ms
Balance Query55ms
Escrow / SettlementNot native
Casino BetNot available

Run Your Own Benchmark

Reproduce our results or measure latency from your own agent's hosting region. The script below benchmarks every public endpoint and prints a summary table.

Python — purpleflea_benchmark.py
#!/usr/bin/env python3
# Purple Flea API Latency Benchmark
# Usage: python3 purpleflea_benchmark.py
# Requires: pip install httpx
import httpx
import time
import statistics
import asyncio

BASE_URL = "https://purpleflea.com/api/v1"
RUNS     = 20

ENDPOINTS = [
    ("GET",  "/health",          None,                "Health Check"),
    ("POST", "/auth/token",       {"agent_id": "test"}, "Auth / Token"),
    ("GET",  "/wallet/balance",  None,                "Wallet Balance"),
    ("GET",  "/trading/markets", None,                "Trading Markets"),
    ("GET",  "/domains/search",  {"q": "agent"},       "Domain Search"),
    ("GET",  "/casino/games",    None,                "Casino Games"),
]

async def measure(client, method, path, body):
    samples = []
    for _ in range(RUNS):
        t0 = time.perf_counter()
        if method == "GET":
            await client.get(BASE_URL + path, params=body)
        else:
            await client.post(BASE_URL + path, json=body)
        samples.append((time.perf_counter() - t0) * 1000)
    return samples

async def main():
    async with httpx.AsyncClient(timeout=10) as client:
        print(f"\nPurple Flea Benchmark ({RUNS} runs each)\n" + "-"*52)
        print(f"{'Endpoint':<22} {'Median':>8} {'P95':>8} {'Min':>8}")
        print("-"*52)
        for method, path, body, label in ENDPOINTS:
            samples = await measure(client, method, path, body)
            med = statistics.median(samples)
            p95 = sorted(samples)[int(0.95 * RUNS)]
            mn  = min(samples)
            print(f"{label:<22} {med:>7.1f}ms {p95:>7.1f}ms {mn:>7.1f}ms")
        print("-"*52)
        print("Done. Share results at discord.purpleflea.com\n")

asyncio.run(main())

purpleflea_benchmark.py

Async benchmark script. Measures median, P95, and minimum latency for Purple Flea endpoints. No API key required for public endpoints.

Language: Python 3.9+ Requires: httpx Samples: 20 per endpoint

How We Measure

Our benchmarks are reproducible and conservative. We measure client-perceived latency, not internal processing time.

📍
Measurement Region
Primary benchmarks from Frankfurt, EU (Hetzner cloud). Secondary from Virginia, US (AWS us-east-1). Published values are the Frankfurt median.
📊
Sample Size
Each published number is the median of 100 consecutive requests. P95 values exclude warm-up outliers. Cold-start first requests are excluded.
🔄
Update Frequency
Benchmarks re-run hourly via our status monitor. The values on this page are updated daily with a rolling 24-hour average.
🔒
Authenticated Endpoints
Trading, casino, escrow, and wallet endpoints are benchmarked with a valid agent token. Auth latency is measured separately.

Fast APIs. Free to Start.

Claim your $1 USDC from the faucet and experience sub-50ms agent finance infrastructure firsthand. No credit card, no KYC.