Health
Healthy
Benchmark
100/100
SDK Version
2.12.1
Audit Status
Audit Done
Started Apr 27, 2026
Approval
Not requested
exa-py is the Exa Python SDK itself. pyproject.toml declares version 2.12.1 which matches the latest published release on PyPI. All declared endpoints are implemented in exa_py/api.py: search (line 1529), stream_search (1673, the streaming variant matching 'search_streaming'), get_contents (1838), find_similar (2004), answer (2454), stream_answer (2536, matching 'answer_streaming'), plus async variants. The research endpoint is implemented via ResearchClient/AsyncResearchClient (exa_py/research/) with create/get/list/poll_until_finished. All seven declared search types are present in SearchType Literal: 'auto', 'fast', 'deep-lite', 'deep', 'deep-reasoning', 'neural', 'instant'. All declared content options are passed through via CONTENTS_OPTIONS_TYPES and CONTENTS_ENDPOINT_OPTIONS_TYPES: text, highlights, summary, livecrawl (with options always/fallback/never/auto/preferred), subpages (plus subpage_target), and extras. Beyond declared capabilities, the SDK also exposes additional features: search_and_contents, find_similar_and_contents, system_prompt, output_schema (DeepTextOutputSchema/DeepObjectOutputSchema), additional_queries (deep variants), context, metadata, livecrawl_timeout, max_age_hours, filter_empty_results, moderation, user_location, category, flags, AsyncExa client, and a websets module. No declared-but-missing capabilities found.