Models · Launch wave B

Use a Custom Model Provider with DeepSeek Harness

A custom provider must match the protocol expected by its configured route, expose a reachable base URL and advertise a usable model. Validate connectivity and the model list before debugging agents, tools or plugins. This page separates verified behavior from community reports and marks the release-candidate context so you can act without mistaking an observation for a permanent guarantee.

Verified Aug 14, 2026Developer previewIndependent guide
Evidence status: Official provider guide. Re-check the linked first-party source after a new release candidate.

What is verified now

A custom provider must match the protocol expected by its configured route, expose a reachable base URL and advertise a usable model. Validate connectivity and the model list before debugging agents, tools or plugins.

Evidence basis: Official provider guide. The status was checked on 14 August 2026; re-check the linked source after any dsh release.

Safe workflow

Use the smallest reversible workflow first. Do not add community plugins or paste credentials while validating the base behavior.

  1. Validate the provider base URL and credential source.
  2. Select an exact model identifier exposed by that route.
  3. Send a minimal request before enabling tools or plugins.

Verify before moving on

A successful result should be observable: a version string, reachable local URL, valid provider response, loaded configuration, or disappearance of the exact error. If the result differs, stop and capture sanitized evidence rather than stacking unrelated fixes.

  1. Record the exact dsh, Node or Python version.
  2. State the operating system and installation method.
  3. Link the first-party source and include only sanitized logs.

First-party sources

Open the original source before acting on a changed version or unresolved community report.

Continue by layer