Skip to main content
v1

Answer collection fixes

  • Fixed ChatGPT collection failures.
  • Fixed Google Overview and Google AI Mode answers being collected truncated.
v1

Added location-level collection

You can now collect answers at the city level, which is narrower than a region (country).
  • Set a location code (for example, US-NEW_YORK) in the location of a run request to collect at that location. If you omit it, collection happens at the region level as before.
  • Retrieve the supported location codes with the newly added List locations API. Use model to filter locations by model, and region to filter by country.
  • Added the location field to the answer retrieval response and the run status response.
  • The platforms that support location are ChatGPT and Perplexity.

Fixed answer collection errors

  • Fixed an issue where Naver AI Briefing answers were occasionally collected truncated partway through.
  • Fixed an issue where Gemini answers were occasionally collected truncated partway through.
v1

Partially resumed Perplexity collection

Answer collection for the PERPLEXITY model, suspended since October 10, is available again.
  • Answers are now collected through Perplexity’s official chat API instead of web scraping.
  • It is disabled by default. Contact Chainshift if you need it.

Fixed answer collection errors

  • Fixed an issue where cited sources (sources) were not collected after the ChatGPT interface changed.
v1

Fixed answer collection errors

  • Fixed an issue where some collections failed on Google Overview.
v1

Suspended Perplexity collection

  • Perplexity tightened its blocking policy on October 10, so answer collection for the PERPLEXITY model is temporarily suspended. Other models work normally.

Fixed answer collection errors

  • Fixed an issue where collection failed intermittently after the ChatGPT response format changed.
  • Fixed an issue where some collections failed on Google Overview.
v1

Fixed ChatGPT answer collection errors

Fixed an issue where new UI elements added to ChatGPT caused unwanted text to appear in collected answer bodies and broke their formatting. There are no changes to the API interface.
v1

Improved answer collection stability

Improved the stability and throughput of ChatGPT and Perplexity answer collection.
  • Fixed ChatGPT encoding errors: resolved save failures for answers containing invalid UTF-8 or NUL characters. The body and source titles are now sanitized before saving
  • Fixed missing ChatGPT citations: resolved an issue where only the representative URL was collected when multiple search results were grouped into a single citation chip (“+N”). All grouped URLs are now reflected in sources and in the inline citation links of the body markdown
  • Guarded against ChatGPT answer contamination: detects the rare case where the original question is saved as content instead of the answer, marks it as failed, and retries
  • Improved Perplexity throughput: reuses collection sessions and improved answer completion detection. Fewer failures from anonymous usage limits and fewer answers cut off mid-sentence
v1

Platform API v1 released

The first public version of the Chainshift Platform API.
  • Collection pipeline: prompt set and prompt management, run requests (8 models, regions, repeated runs), answer collection (body, cited sources, fan-out queries)
  • Statistics: cited domain and URL rankings, fan-out query aggregation
  • Keywords: Google and Naver search volume metrics, related keywords, autocomplete suggestions, Naver trends, site keyword analysis
  • Brands: management of the brands used as the basis for analysis (name, synonyms, website)
  • MCP: an MCP server mapped to the main APIs (/mcp)
Backward compatibility is maintained within the v1 path, and every change is recorded on this page. For the full policy, see Versioning and compatibility.