pixserp

pixserp is an AI-native search API built by TetiAI LLC for developers who need live web data inside agents, RAG pipelines, and research tools. It exposes a single /v1/chat/completions endpoint that is drop-in compatible with the OpenAI SDK, returning cited, structured answers instead of raw HTML. The service launched with a pay-as-you-go model and offers a free starting balance with no credit card required.

What It Is

pixserp sits between an LLM application and the live web, handling search, scraping, and synthesis in one flat-rate API call. Rather than returning a wall of HTML for the developer to parse, it delivers a cited answer plus structured fields — product listings, flight results, hotel cards, video transcripts — depending on which of its ten "answer shapes" the query triggers. The vendor positions it as a replacement for the naive RAG pattern of calling a search API, scraping pages, and feeding thousands of tokens to a language model.

Ten Answer Shapes, One Endpoint

The core differentiator pixserp advertises is breadth of structured output from a single endpoint:

• Web — cited article-style answers
• News — recent coverage with sources
• Images — reference image results
• Places & Maps — local business results
• Shopping — product listings with price filters
• Flights — cheapest direct routes
• Hotels — availability and star-rating filters
• YouTube videos — relevant video results
• YouTube transcripts — summarization of a given video URL
• Any URL — key-claim extraction from an arbitrary page

The vendor claims competitors such as Perplexity Sonar, Exa /answer, and Brave Summarizer cover only a subset of these shapes.

Research Depth Modes

pixserp offers four model variants that trade latency for thoroughness:

• pixserp-fast — quick cited answer, lowest latency (the vendor cites ~1.5 s)
• pixserp-standard — balanced research with key-fact verification
• pixserp-deep — thorough cross-referenced research
• pixserp-agent — multi-step research loop billed per step rather than per request

All four share the same OpenAI-compatible /v1/chat/completions interface and support JSON schema outputs and streaming SSE.

Drop-In OpenAI SDK Compatibility

Developers point the base_url of the official openai Python (or other language) SDK at https://pixserp.com/v1 and swap the API key. Existing chat-completions code continues to work; citations come back as a structured field on the message object. The vendor also ships an MCP server for agent frameworks that support the Model Context Protocol.

Cost Model and Positioning

pixserp uses flat per-request pricing with no token bill on top and no subscription tiers. Top-ups range from $50 to $50,000 and balances never expire. The vendor publishes a cost comparison (sourced from public list prices as of May 2026) claiming pixserp-fast is cheaper per 1,000 cited-answer requests than Exa /answer, Brave Answers, and Perplexity Sonar Pro. Higher-concurrency arrangements are available by contacting support.

Current Status

The product is live and accepting sign-ups via magic-link with a free starting balance. The homepage references a Product Hunt listing (post ID 1148331) and the copyright footer reads "© 2026 TetiAI LLC." The MCP server page and affiliate program are also active. No version number or changelog is publicly surfaced in the available sources.