# Can I use an agent without a local model?

A local model is one provider option. Separate model access from the assistant runtime, and keep a non-model fallback for collection tasks.

- Citation: https://botclaw.tech/items/obs-topic-model-access
- Revision: 2
- Published: 2026-10-01T16:24:26.193Z
- Reviewed: 2026-10-01T15:50:00.000Z
- Updated: 2026-10-01T16:24:32.310Z
- Section: topic
- Underlying source date: Unknown; do not infer a release date
- Review state: current; 0 pending source changes
- Applicable version: Not pinned; check current documentation

The first decision is whether a task needs inference at all. Fetching feeds, deduplicating URLs and rendering a list can be ordinary software work.

## Separate three dependencies

The runtime hosts tools and state. A model provider supplies inference. A channel delivers messages. A healthy channel or an installed runtime does not prove that the model endpoint is reachable.


## Choose an honest starting point

Use a compatible remote provider when it fits your data and budget requirements; use a local model only if a reachable service is actually available. Without either, build deterministic collection and source review first.


## Acceptance sequence

1. Check endpoint reachability separately from a real authorized inference request.
2. Run one fixed small task, record the output and cost, then add a tool.
3. Retain a plain collector path when inference is unavailable.

## Evidence and limits

- official_documentation
- Source-backed facts plus BotClaw editorial interpretation. No product account, inference or agent performance test was run.

## Sources

- https://docs.openclaw.ai/start/openclaw
- https://hermes-agent.nousresearch.com/docs/getting-started/quickstart