diff --git a/agents/openclaw-main/warp-acp-ops.md b/agents/openclaw-main/warp-acp-ops.md
index c01501b..b686516 100644
--- a/agents/openclaw-main/warp-acp-ops.md
+++ b/agents/openclaw-main/warp-acp-ops.md
@@ -58,6 +58,34 @@ Each wrapper should:
 3. fix provider + model mapping
 4. run `opencode-ai acp`
 
+## Active Default Routing Baseline (2026-03-18)
+
+Current `opencode` default-model routing baseline:
+
+- `mac-5`
+  - default model: `opencode/minimax-m2.5-free`
+  - rationale: the minimax free cloud route already passed direct + ACP minimal validation and avoids the timeout-prone local `oMLX` ACP path
+- `mac-6`
+  - provider label: `vllm`
+  - base URL: `http://100.64.0.5:8000/v1`
+  - model id: `Qwen3.5-27B`
+  - default model string: `vllm/Qwen3.5-27B`
+- `mac-7`
+  - provider label: `vllm`
+  - base URL: `http://100.64.0.5:8000/v1`
+  - model id: `Qwen3.5-27B`
+  - default model string: `vllm/Qwen3.5-27B`
+
+Superseded assumption:
+
+- Do not continue to describe the cluster rule as "each machine defaults to its own local model" for `opencode` ACP.
+- That older description was true for the earlier `oMLX`-first phase but is no longer the active default-policy baseline.
+
+Validation snapshot:
+
+- `mac-6` ACP test result: `MAC6_VLLM_ACP_OK`
+- `mac-7` ACP test result: `MAC7_VLLM_ACP_OK`
+
 ## Key Files
 
 ### OpenClaw
diff --git a/daily/2026-03-17.md b/daily/2026-03-17.md
index 5188241..6e09178 100644
--- a/daily/2026-03-17.md
+++ b/daily/2026-03-17.md
@@ -72,3 +72,23 @@
   - `sessions_spawn(runtime="subagent")` 在 `local-mac5/local-mac6/local-mac7` 上的真实调度行为
   - `nodes run -> harness/opencode/acpx` 超时根因
   - gateway 重启后 memory / context engine 新配置的实际运行行为
+
+## 2026-03-18 凌晨补充：opencode 默认模型规则更新
+
+- 原先可简写为“默认使用本机本地模型”的 `opencode` 规则已不再准确，现已按机器角色重新分流
+- `mac-5` 当前 `opencode` 默认模型改为：
+  - `opencode/minimax-m2.5-free`
+- 这样做的原因是：此前 direct `opencode` 与 ACP 最小测试都已验证 minimax free 路径可用，且它能绕开本地 `oMLX` 在 ACP/session 路径上的超时脆弱点
+- `mac-6` 与 `mac-7` 当前 `opencode` 默认模型改为共享 vLLM 接口：
+  - base URL: `http://100.64.0.5:8000/v1`
+  - model id: `Qwen3.5-27B`
+  - provider/model 字符串：`vllm/Qwen3.5-27B`
+- 已从 `mac-6` 和 `mac-7` 实测 `GET http://100.64.0.5:8000/v1/models`，当前暴露模型为：
+  - `Qwen3.5-27B`
+- 已将两台机器的 `~/.config/opencode/opencode.json` 切到上述 vLLM 配置，并完成 ACP 最小测试：
+  - `mac-6` -> `MAC6_VLLM_ACP_OK`
+  - `mac-7` -> `MAC7_VLLM_ACP_OK`
+- 因此当前稳定默认规则应记为：
+  - `mac-5`：`opencode/minimax-m2.5-free`
+  - `mac-6`：`vllm/Qwen3.5-27B` via `http://100.64.0.5:8000/v1`
+  - `mac-7`：`vllm/Qwen3.5-27B` via `http://100.64.0.5:8000/v1`
diff --git a/daily/2026-03-18.md b/daily/2026-03-18.md
new file mode 100644
index 0000000..57cd8bc
--- /dev/null
+++ b/daily/2026-03-18.md
@@ -0,0 +1,23 @@
+# 2026-03-18
+
+## opencode 默认模型规则更新
+
+- 原先“`opencode` 默认使用各机器本地模型”的描述已失效，不应继续作为当前默认规则使用
+- 当前生效的默认策略为：
+  - `mac-5`：`opencode/minimax-m2.5-free`
+  - `mac-6`：`vllm/Qwen3.5-27B` via `http://100.64.0.5:8000/v1`
+  - `mac-7`：`vllm/Qwen3.5-27B` via `http://100.64.0.5:8000/v1`
+
+## 验证记录
+
+- `mac-5` 上 direct `opencode` 与 ACP 最小测试已通过 `opencode/minimax-m2.5-free`
+- 从 `mac-6` / `mac-7` 实测 `http://100.64.0.5:8000/v1/models`，可见模型：
+  - `Qwen3.5-27B`
+- 已将 `mac-6` / `mac-7` 的 `~/.config/opencode/opencode.json` 更新为 `vllm` provider 配置
+- `mac-6` ACP 最小测试结果：`MAC6_VLLM_ACP_OK`
+- `mac-7` ACP 最小测试结果：`MAC7_VLLM_ACP_OK`
+
+## 备注
+
+- 这次更新针对的是 `opencode` 默认模型策略，不等于废弃原有 subagent / worker 的本地模型拓扑说明
+- 后续涉及 `opencode` 默认模型时，应优先引用本日规则，而不是旧的“每机默认本地 oMLX”说法
diff --git a/shared/long-term/projects/warp-acp-integration.md b/shared/long-term/projects/warp-acp-integration.md
index 64f9877..1ea8302 100644
--- a/shared/long-term/projects/warp-acp-integration.md
+++ b/shared/long-term/projects/warp-acp-integration.md
@@ -87,6 +87,28 @@ Future examples:
 - `gemini-warp-infini-kimi`
 - `claude-warp-ckimi-kimi`
 
+## Current Default Model Policy (2026-03-18)
+
+The cluster no longer uses a single "default to each machine's own local model" rule for `opencode` ACP.
+
+Current node-specific default policy:
+
+- `mac-5`: default `opencode` model is `opencode/minimax-m2.5-free`
+- `mac-6`: default `opencode` model is `vllm/Qwen3.5-27B` via `http://100.64.0.5:8000/v1`
+- `mac-7`: default `opencode` model is `vllm/Qwen3.5-27B` via `http://100.64.0.5:8000/v1`
+
+Operational meaning:
+
+- `mac-5` prefers the already-validated free cloud minimax route for daily ACP stability.
+- `mac-6` and `mac-7` prefer the shared local vLLM endpoint instead of their previous per-node local `oMLX` default for `opencode` ACP.
+- This rule is specific to current `opencode` defaults; it does not invalidate separate worker/subagent topology docs.
+
+Observed validation status:
+
+- `mac-5`: direct `opencode` and ACP minimal tests succeeded with `opencode/minimax-m2.5-free`
+- `mac-6`: ACP minimal test succeeded with `vllm/Qwen3.5-27B`
+- `mac-7`: ACP minimal test succeeded with `vllm/Qwen3.5-27B`
+
 ## Configuration Layer Responsibilities
 
 ### OpenClaw