We don't publish
your competitive advantage.

IFCalling `bind_tools()` on a `ChatOllama` instance raises `AttributeError: 'ChatOllama' object has no attribute 'bind_tools'`.

THENUse a chat model that natively supports `bind_tools` (e.g., `ChatOpenAI`) or switch to `OllamaFunctions` from `langchain_experimental` for experimental tool calling. Alternatively, wait for a future release of `langchain-core` where `ChatOllama` may implement `bind_tools`.

IFUsing create_sql_agent with agent_type='tool-calling' and ChatMistralAI causes HTTP 400 error: 'Expected last role User or Tool (or Assistant with prefix True) for serving but got assistant'.

THENWhen using Mistral AI models with tool-calling site_1, avoid the error by switching to a different agent type such as 'zero-shot-react-description', which does not inject assistant messages before tool calls. Alternatively, customize the agent prompt to instruct the model to output tool calls directly without intermediate reasoning. The issue stems from LangChain's tool-calling agent violating Mistral's requirement that the last message before a tool invocation be from the user or tool role.

IFCalling llm.bind_tools() on a ChatOllama instance raises AttributeError: 'ChatOllama' object has no attribute 'bind_tools'.

THENChatOllama does not implement bind_tools. To use tool calling with Ollama models, switch to OllamaFunctions from langchain_experimental or use a different chat model (e.g., ChatOpenAI, ChatMistral) that supports bind_tools. Alternatively, wait for an update to langchain-core (>=0.1.41) which will provide a clearer error, but the feature may still not be available until ChatOllama implements it.

IFMistral Small 3.1 in HF format fails to start with `--tokenizer-mode mistral` when using `--enable-auto-tool-choice`

THENAvoid the built-in mistral tokenizer; instead provide a custom chat template via `--chat-template` that includes the proper Jinja2 template for Mistral tool calling. This bypasses the `MistralTokenizer` compatibility issue in vLLM v0.8.3/v0.8.4.

IFUsing Python's @tool decorator with OllamaFunctions in create_tool_calling_agent results in a JSON serialization error during agent invocation.

THENEnsure the language model supports native tool calling. For OllamaFunctions, use a model variant that supports function calling or switch to a model like OpenAI that has proven compatibility. Alternatively, manually bind tools using model.bind_tools() instead of using the agent constructor.