PulseAugur
EN
LIVE 18:34:17

Perplexity AI launches "Search as Code" for agents

By PulseAugur Editorial · [6 sources] ·

Perplexity AI has introduced "Search as Code," a novel search architecture designed for AI agents. This new system bypasses traditional, high-latency tool-calling methods by allowing models to directly compose search primitives, enabling asynchronous query execution, deduplication, and result ranking. Perplexity claims Search as Code outperforms existing systems on various deep and wide research benchmarks, including their own WANDR benchmark, while also offering a superior cost-performance ratio. AI

IMPACT This new architecture could significantly improve AI agent efficiency and performance in research tasks by reducing latency and enhancing search capabilities.

RANK_REASON Product launch of a novel AI architecture with benchmark performance claims.

Read on X — Perplexity →

AI-generated summary · Google Gemini · from 6 sources. How we write summaries →

Perplexity AI launches "Search as Code" for agents

COVERAGE [6]

  1. X — Perplexity TIER_1 English(EN) · perplexity_ai ·

    More info about Search as Code in the Perplexity Agent API docs:

    More info about Search as Code in the Perplexity Agent API docs: https://t.co/iaXcFSmiWh

  2. X — Perplexity TIER_1 English(EN) · perplexity_ai ·

    WANDR is our in-house wide benchmark, built to mirror real professional research workloads.

    WANDR is our in-house wide benchmark, built to mirror real professional research workloads. Search as Code scores 0.386 to the next best system's 0.152, and the benchmark is far from saturated. We're releasing WANDR in the coming weeks. https://t.co/AkxqOdisaA

  3. X — Perplexity TIER_1 English(EN) · perplexity_ai ·

    It also sets a new cost-performance frontier.

    It also sets a new cost-performance frontier. On DSQA it scores 0.871, ahead of Anthropic's 0.815, at nearly half the cost per task. On WideSearch it leads on score while running cheaper. https://t.co/zG0LuKrWoX

  4. X — Perplexity TIER_1 English(EN) · perplexity_ai ·

    We tested Search as Code on deep research (DSQA, BrowseComp, HLE) and wide research benchmarks (WideSearch, WANDR).

    We tested Search as Code on deep research (DSQA, BrowseComp, HLE) and wide research benchmarks (WideSearch, WANDR). It matches or beats every competing system across all five. https://t.co/tdZhmn6HhD

  5. X — Perplexity TIER_1 English(EN) · perplexity_ai ·

    @PPLXDevs The traditional tool-calling approach suffers from high latency, manual control flow, and context pollution.

    @PPLXDevs The traditional tool-calling approach suffers from high latency, manual control flow, and context pollution. With Search as Code, the model composes search primitives: fanning out queries asynchronously, deduping, filtering, joining, and ranking before results hits its…

  6. X — Perplexity TIER_1 English(EN) · perplexity_ai ·

    Introducing Search as Code, our new search architecture for AI agents.

    Introducing Search as Code, our new search architecture for AI agents. It writes Python that calls our search stack directly, instead of looping through function calls one at a time. Available in the Perplexity Agent API, and now default in Computer. https://t.co/ut6GGWQTVO ht…

RELATED ENTITIES

RELATED TOPICS