Problemsby GlyphTrouble with Anthropic messages
4 posts from 4 people4d activeSteady, +1 this week
Posts per day
The posts behind it
[Bug]: /v1/messages: client disconnect before the first chunk leaves the upstream LLM request running
## What happened? LiteLLM 1.106.2 proxy, a non-Anthropic model served through `/v1/messages` (`custom_openai/...` → OpenAI-compatible backend, `stream: true`). When the client disconnects while the proxy is still waiting for the first chunk, `create_response` logs but the...
convert_to_openai_messages drops invalid tool calls from exported conversations
### Submission checklist - [x] This is a bug, not a usage question. - [x] I added a clear and descriptive title that summarizes this issue. - [x] I used the GitHub search to find a similar question and didn't find it. - [x] I am sure that this is a bug in LangChain rather than...
Anthropic and OpenAI Responses formatters stringify multimodal parts content in system/tool messages (follow-up to #7920)
### Description Follow-up to #7920 (Bedrock Converse, fixed by #7930): `LLMMessage.content` is `str | list[dict[str, Any]] | None`, and four more formatter sites stringify the parts form with an f-string or `str()`, putting the list repr into API-bound fields. AGENTS.md's...
[Bug]: concurrent /v1/messages calls sometimes answer 500 "dictionary changed size during iteration" while the spend row records a success
### Description Under concurrent load, a non-streamed `POST /v1/messages` that the provider served fine occasionally answers `500 {"error": {"message": "dictionary changed size during iteration", ...}}`. The spend row still lands with the right cost, so the user is billed for a...
Companies and products named
About this problem
Evidence
4 posts from 4 people in 3 places, about 1 a week over 19 days. Mostly on BerriAI/litellm, crewAIInc/crewAI, langchain-ai/langchain. Tools named alongside: OpenAI, Gemini, Text, Docker.
Frustration Frustration 2 of 3· Seen on GitHub issues
How it was grouped
Posts about Anthropic with no known issue, sharing the word "messages". First post Oct 6, 2026, latest Oct 24, 2026. Corroborated: 3 or more posts from 2 or more people or places. Method
Other Anthropic problems
Anthropic upgrades break working setups
26 posts from 21 people on GitHub issues.
Anthropic rate limits and usage caps block work
12 posts from 9 people on GitHub issues, Reddit, Community forums.
Anthropic users keep filing bugs
Top post: "`HTMLHeaderTextSplitter` raises `ValueError` on init for the non-heading tags its splitter already supports", one of 31 posts from 30 people on GitHub issues, App Store reviews, Community forums.
Trouble with Anthropic core
Top post: "core: filter_messages(exclude_tool_calls=[...]) does not remove invalid_tool_calls", one of 6 posts from 5 people on GitHub issues.
Trouble with Anthropic crashes
Top post: "Streaming parser crashes when a tool-call delta has function=None", one of 4 posts from 4 people on GitHub issues.
Anthropic token costs are hard to predict
4 posts from 4 people on GitHub issues, Hacker News, Reddit.
RAG on Anthropic returns irrelevant or missing context
4 posts from 4 people on GitHub issues, Reddit.
Anthropic fails to return valid structured output
3 posts from 3 people on GitHub issues.
Trouble with Anthropic inside
Top post: "Agentic run with dynamic filtering (web_search_20260209 inside code_execution) rejected with code_execution", one of 3 posts from 3 people on GitHub issues.
Trouble with Anthropic instead
Top post: "Tool `config: RunnableConfig` parameter receives the parent config instead of the child config, so nested", one of 3 posts from 3 people on GitHub issues.
Anthropic lock-in makes switching costly
3 posts from 3 people on GitHub issues, Hacker News.
Prompt injection makes Anthropic features unsafe
3 posts from 3 people on GitHub issues, Hacker News.
History