Textkernel Tx Platform October Release (2026.10)
What’s in this release
Tx Platform caps LLM parsing file size, restores Flex Requests for non-LLM parsing, adds Japanese and Hebrew CV parsing in the Australian data center, tightens certification extraction, and surfaces clearer SDK error messages.
LLM Parsing Maximum File Size
LLM parsing requests are now capped at a maximum file size of 3MB.
FixedCV and Resume Parsing Timeout Errors
Timeout errors are now reported clearly instead of surfacing as unhandled exceptions.
FixedCertification Section Recognition
Certification extraction now adheres to section boundaries more accurately.
FixedFlex Requests for Non-LLM Parsing
The Flex Requests feature now works again with standard non-LLM parsing requests.
FixedJapanese and Hebrew CV Parsing in the Australian Data Center
Japanese and Hebrew CVs now parse correctly in the Australian data center.
FixedTX Platform Match V1 Index Creation Error Fix
Creating a Match v1 index on TX Platform no longer occasionally fails with an incorrect 409 error.
FixedSuggest-Skills-From-Skills Endpoint Response Fix
The suggest-skills-from-skills endpoint now returns a 200 with an empty list, instead of a 404, when a valid skill code has no suggestions.
FixedTx Platform Match V1 Deprecation
TX Platform Match v1 (formerly Sovren Match) is now deprecated. New implementations should use Match v2.
FixedFast Parser Region Field Normalization Regression
The fast parser now correctly returns the unnormalized region value, reverting a regression that normalized it unintentionally.