LangChain: Decisions API Beta and Agent Reliability Fixes
LangChain shipped a beta wrapper for OpenAI's Decisions API alongside OpenAI, core, and LangChain releases. The broader pattern is reliability — smarter retries and fixes for tool-call conversion, token counting, and Bedrock system messages.
Duration: PT2M26S
Episode overview
This episode is a short developer briefing from LangChain.
It explains recent repository work in plain language.
- Show: LangChain
- Published: 2026-10-09T13:01:59Z
- Audio duration: PT2M26S
Transcript excerpt
This excerpt keeps the crawler page concise. Listen to the episode or use the RSS feed for the full update.
Good morning, it's Friday, October 9th, 2026, and here's your LangChain briefing.
The lead signal is resilience for long-running agents: a major new integration landed while core tightened retry logic, summarization, and message handling.
First, new capability. Pull request 41125 adds beta support for OpenAI's Decisions API, released as part of OpenAI version 1.7.0. It supports LangChain message types, includes built-in LangSmith tracing, and ships prebuilt middleware matching the TypeSafe integration. There's deliberately no shared base class in…
Second, staying alive under failure. Pull request 41158 lets the retry helper accept a callable to decide if an exception is retryable, useful for checking the new standard model errors. Building on that, pull request 41159 fixes the summarization middleware to retry its summary step with conservative trimming when…
Third, message fidelity. Three open fixes target silent data loss. One ensures invalid tool calls with unparsable arguments are included in OpenAI message conversion, preventing orphaned tool replies that providers reject. Another corrects the approximate token counter to include tool calls hidden in list-form…
What's next: test…
Nearby episodes from LangChain
- Tool Message Fixes and Mid-Conversation Releases
- Weekly Recap - Profiles, Decisions, and Message Reliability
- Safer Agent Sandboxing and Measurement Fixes
- Mid-Conversation Profiles and Splitter Fixes
- Reasoning Fixes and Decisions API Beta
- Streaming Fidelity and Core Edge Cases
- Sharper Agent Guardrails and Credential Hardening
- Credential Leak Fixes and Required OpenAI Model