ApePredictor API Documentation
BAYC valuations from the accepted model, verified sales, and a unified prediction-before-sale performance record.
Valuations are scheduled every two hours. Failed updates leave the last approved values visible. Check freshness metadata: HTTP 200 does not guarantee fresh data. Estimates are not guaranteed sale prices.
https://www.apepredictor.xyzAccess
Valuations, valuation history, batches, and deals require an active paid membership. API clients send their member key in the X-API-Key header or Authorization: Bearer <member-key>. Keep keys out of URLs. Existing members sign in at Access; browser requests use a secure HttpOnly session whose membership is checked on every request. See membership plans for pricing and availability. Existing members can renew from their Account panel.
Listings, realized sales, floors, and historical performance remain public. Missing, expired, or revoked membership returns HTTP 401; a membership service outage returns HTTP 503.
Collector memberships delay access to newly listed Apes by 90 minutes. Direct valuations and valuation history return HTTP 403 with code collector_listing_delay during that window; batches and deal lists omit those Apes. Full Access has no listing delay.
Administrative mutations are not available through the public frontend.
Request bounds
Batches accept at most 25 unique token IDs (0–9999). Reuse results, keep concurrency low, and back off on 429 or 503 responses. No per-key rate quota or public API SLA is currently promised.
Read endpoints
Current valuation and uncertainty
{"nft_identifier":123}Includes the published estimate, supported intervals, state, stale, and source_data_asof. Intervals are not neural-network ensemble members.
Ape historical valuations
Returns up to 100 historical points, oldest to newest. Retention preserves daily snapshots and scored-sale evidence.
Bounded valuation batch
{"nfts":[{"nft_id":123},{"nft_id":456}]}Results are keyed by token ID. Missing estimates remain null; never interpret them as zero. Fetch collections in bounded batches, not one 10,000-token request.