pixserp
One API for the entire web·pixserp.com
pixserp is an OpenAI-compatible AI search API that returns cited, structured answers from the live web through a single endpoint. It covers ten answer shapes — web, news, images, places, shopping, flights, hotels, YouTube, transcripts, and any URL — via drop-in models named pixserp-fast, pixserp-standard, pixserp-deep, and pixserp-agent. It also offers a scheduled 'Watch' mode that reruns queries on a cron and fires webhooks on change, plus an MCP server for tool integration. Pricing is flat, pay-as-you-go per request/step, with no subscription tiers.
What it's for
Features 15
- API key management from dashboard (SHA-256 hashed storage)
- Diff-aware, HMAC-signed webhooks showing added/removed/modified entries
- Drop-in replacement for the openai SDK (Python, JavaScript, Go, curl)
- Four models: pixserp-fast, pixserp-standard, pixserp-deep, pixserp-agent
- Inline [n] citations with structured citation objects per answer shape
- JSON schema output support via response_format
- MCP server with streamable HTTP transport for tool use in Claude Desktop, Cursor, Zed, Cline, Continue
- Model discovery endpoint (/v1/models)
- Multi-step research agent (pixserp-agent) that searches, scrapes, follows links, and cross-checks
- OpenAI-compatible /v1/chat/completions and /v1/responses endpoints
- Per-type signal whitelist to filter noise in diffs
- Per-user rate-limit pool shared across endpoints
- Server-Sent Events (SSE) streaming with sub-second time-to-first-token
- Ten answer shapes: web, news, images, places, shopping, flights, hotels, YouTube, transcripts, any URL
- Watch: scheduled cron-based reruns (1-minute to daily) of any chat completion
At a glance
Integrations
Resources
Pricing
Pay-as-you-go pricing; top up balance from $50 up to $50,000, balance never expires. New accounts get $2.50 free credit to start, no card required.
Quick cited answer, Minimal latency, 1 search
Flat per-request pricing, no tiers
Balanced research, Verified key facts
Flat per-request pricing, no tiers
Thorough cross-referenced research, Multi-angle verification
Flat per-request pricing, no tiers
Orchestration in one call replacing a full retrieval pipeline
Bills only steps actually run
Re-runs any chat completion on a cron and fires a webhook on diff
Cadence floor 1 minute; cron and webhook are free