Market Orders: rate limit rolls out on February 24, 2026
Hi all,
We're continuing to put more ESI routes under rate limiting. This time it's the turn of the market-order group. The only route in this group is /markets/{region_id}/orders — and it's a big one.
Why this route?
The original plan was to put all market-related routes into a single "market" group. That turned out to be really difficult to do in a sensible way, so we split things up. The market-order group is the first piece to be rolled out.
The main reason we're doing this now: this route is the most common one left not under rate limiting that still causes bursts of traffic. In practice, there are two repeating patterns: someone has a bug in their script, or they didn't notice the route is cached for 5 minutes and they're polling it more often than that. Both lead to a lot of unnecessary load on ESI and on Tranquility. Nobody wins.
With a clear limit in place, we're hoping everyone can stay on the right side of the line and avoid getting bonked by the banhammer.
You can find the full details on how rate limiting works in our rate limiting documentation.
Tokens
We're giving this group the largest amount of tokens yet: 12,000 tokens. That's enough to fetch every region's orders every 5 minutes if you really wanted to. This is not an invitation to do that; please only fetch the data you're actually using.
For those interested in the napkin math: across all 113 regions visible in ESI, there are 1,723 pages worth of market orders (at the time of writing this post). Each page is cached for 5 minutes, so each can be fetched 3 times within the rate limit window. This means that fetching every region’s orders and updating it as the cache expires would consume 10,338 tokens (1,723 pages × 2 tokens per request × 3 requests per window). That’s well within the proposed 12,000 tokens.
When and who
Rate limiting for the market-order group will be enabled on 24 February 2026.
Some applications will be impacted. The ones we've seen that would exceed the limit use non-descriptive User-Agents, such as Python/3.13 aiohttp/3.12.15 or akka-http/10.2.9. If that sounds like your app, please give your client a descriptive User-Agent and double-check how often you call this route. Respect the 5-minute cache — there's no benefit to polling faster.
Wrapping up
For most applications, this change will be invisible. The limit is only there to rein in scripts that do more than they're likely meant to, whether due to a bug or by polling faster than the cache makes useful. If you're already respecting the cache and only requesting the regions you need, you're in good shape.
Cheers,
Your friendly developers of the EVE Speed Inhibitor (working title)