openclaw - 💡(How to fix) Fix OpenCode Go provider: deepseek-v4-pro and deepseek-v4-flash missing from model list [1 comments, 2 participants]

Description

When configuring OpenCode Go as a provider through OpenClaw's built-in config UI, the model list does not include deepseek-v4-pro or deepseek-v4-flash, despite these models being available on the OpenCode Go API.

Steps to Reproduce

1. Open OpenClaw config
2. In the model/provider section, select OpenCode Go from the provider list
3. Enter a valid OpenCode Go API key
4. View the available model list
5. Observe that deepseek-v4-pro and deepseek-v4-flash are not listed

Expected Behavior

deepseek-v4-pro and deepseek-v4-flash should appear in the model selection list, as they are available on the OpenCode Go API.

Evidence

The OpenCode Go /v1/models endpoint returns these models (confirmed via GET https://opencode.ai/zen/go/v1/models):

• deepseek-v4-pro (owned_by: opencode)
• deepseek-v4-flash (owned_by: opencode)

Full model list from the endpoint: minimax-m2.7, minimax-m2.5, kimi-k2.6, kimi-k2.5, glm-5.1, glm-5, deepseek-v4-pro, deepseek-v4-flash, qwen3.6-plus, qwen3.5-plus, mimo-v2-pro, mimo-v2-omni, mimo-v2.5-pro, mimo-v2.5

Possible Cause

• OpenClaw may be using a hardcoded/cached model list for the OpenCode Go provider that was built before deepseek-v4 was added
• The provider plugin may be filtering models by a known-IDs allowlist instead of dynamically fetching from the API

Environment

• Provider: OpenCode Go (configured via OpenClaw config UI)