LangChain shipped langchain==1.4.3 on 2026-09-28 (on PyPI): init_chat_model supports AWS Bedrock Mantle chat models via langchain-aws Mantle interfaces; recognizes GPT-6 structured output without profiles; create_agent emits error ToolMessages for invalid tool calls missing results on replay; sanitizes cache settings for fallback models. Key PRs: #40837, #40844, #40530, #40886 vs 1.4.2.

Key Takeaways

  • ✓langchain==1.4.3 on PyPI (2026-09-28)
  • ✓init_chat_model + Bedrock Mantle (#40837) via langchain-aws Mantle chat wrappers
  • ✓GPT-6 structured output recognized without profiles (#40844)
  • ✓create_agent repairs invalid tool calls with error ToolMessages (#40530); sanitize fallback cache settings (#40886)
🧭

Finished reading? Explore benchmark rankings & pricing

Real-world SWE-bench scores & $20/mo vs API cost break-even calculator

🔬

In-Depth Technical Analysis

Core Background & Industry Pain Points Agent stacks must absorb new vendor entry points and dirty tool-call histories: Bedrock Mantle needs first-class init_chat_model support; GPT-6 structured output shouldn’t require profiles; create_agent must survive invalid tool calls missing results on replay. ### Architecture Highlights & Internals [langchain==1.4.3](https://github.com/langchain-ai/langchain/releases/tag/langchain%3D%3D1.4.3): #40837 Bedrock Mantle in init_chat_model (langchain-aws Mantle wrappers); #40844 GPT-6 structured output without profiles; #40530 error ToolMessages for invalid tool calls; #40886 sanitize fallback cache settings. ### Authoritative Benchmarks & Measured Scores No public SWE-bench table in this point release—validate Mantle init, GPT-6 SO, and tool-call repair on your agents. ### Developer Hands-on Guide pip install -U "langchain==1.4.3" (+ langchain-aws ≥1.7.9 for Mantle). See docs, PyPI, release.