AlternativesSourced from Prediction market API rate limits, tested with a daily ramp
Kalshi alternatives
CFTC-regulated event contracts exchange with a token bucket API
Kalshi runs a CFTC-regulated event contracts exchange with the only documented token bucket in its cohort (around 20 requests per second on the basic read tier at the time of writing). Its order book and single-market endpoints go to origin on every request, but the market list is served from CloudFront with `max-age=15`, so list latency measures the edge rather than the API origin. The WebSocket requires authentication, which excludes Kalshi from the public WS panel. Builders sizing against this kind of API need two numbers: the real latency of the hot endpoints at a polite request rate, and what happens as the rate climbs. This benchmark probes book (`/markets/{ticker}/orderbook`), price (`/markets/{ticker}`) and list (`/markets`) endpoints continuously from three Railway regions (us-east, eu-west, sgp), every 5 to 7 seconds over warm keep-alive connections, with a separate cold-connect probe once per minute. Once per day per venue it runs a 60-second-per-tier ramp at rising request rates, recording added latency against the same-hour warm baseline and any 429 onset. Kalshi's ramp uses tiers 25/50/100 requests per 10 seconds and stops at the first 429 out of respect for the documented contract. The harness flags CDN cache hits via `cf-cache-status`, `x-cache` and `age` headers and excludes them from latency aggregates so no venue ranks on its CDN. Every request carries an identifying User-Agent with a contact address. The cohort competing with Kalshi includes Polymarket (Cloudflare-queued CLOB), Limitless (undocumented limits, errors CDN cached for 4 hours), Manifold (500 req/min per IP, whole API behind a 5-second cache) and Myriad (single-region Heroku origin, no order book).
About Kalshi: kalshi.com
Top 4 Kalshi alternatives
- Best
- 94 ms
- Median
- 129 ms
- Worst
- 243 ms
- Spread
- 9.5×248 ms → 2.35 s
- Samples · 24h
- 244,1555 providers
Warm book latency · last 24 hours
Warm book latency · last 24 hours
Product ledger · sorted by p50
| Product | Warm book p50 | Trend | |||||||
|---|---|---|---|---|---|---|---|---|---|
| № | Name | Warm book p50 | 24h | ||||||
| 01 | 94 ms | ||||||||
| 02 | 124 ms | ||||||||
| 03 | 129 ms | ||||||||
| 04 | 206 ms | ||||||||
| 05 | 243 ms | ||||||||
Same data as /benchmarks/pm-rate-limits, refreshed every minute. OpenChainBench is community-run; methodology is open.