**TL;DR:** init_chat_model from langchain.chat_models is the unified way to build a chat model. You can pass a bare model name and let it infer the provider, or use a provider:model string to be explicit.

## The fix

1. Prefer init_chat_model for generic code and create_agent `model=` arguments. Model strings like "openai:gpt-5.5" are what the docs use in agent examples.

2. Reach for the provider class when you need provider-only parameters (for example ChatOpenAI's use_responses_api, which picks the Responses vs Completions API).

3. For OpenAI-compatible endpoints (Together, vLLM), pass `model_provider="openai"` with a `base_url`. For routers like OpenRouter or LiteLLM, use the dedicated integrations instead: ChatOpenRouter from langchain-openrouter, or ChatLiteLLM / ChatLiteLLMRouter from langchain-litellm.

## When to use this

- This covers exactly what the title says: init_chat_model and model strings.
- You are setting this up for the first time, or auditing an existing setup.
- You want the key gotchas in one place before you start.

## When not to use this

- You are doing a different workflow with init_chat_model; these steps are specific to the title above.
- You need the full reference docs; this is the short path, not the manual.

## Compatibility

- Mentions 5.5 in the original; behavior may differ on much older or newer releases.

---

## Vectle citations

- Exact version: <https://vectle.com/skills/skl_6RoOnxzHAX_ZzDd3L9W4Gg/versions/skv_CJMzC86jqPNodpLC1gmzlA>
- Canonical HTML skill page: <https://vectle.com/skills/skl_6RoOnxzHAX_ZzDd3L9W4Gg>
- Current Markdown URL: <https://vectle.com/skills/skl_6RoOnxzHAX_ZzDd3L9W4Gg/markdown>

## Selected public attribution and original notices/references

```json
{
  "author": "issueatlas",
  "client": null,
  "notices_and_references": [
    "https://docs.langchain.com/oss/python/langchain/models",
    "https://docs.langchain.com/oss/python/langchain/quickstart"
  ]
}
```