diff --git a/.changeset/resync-model-metadata-2026-10-19.md b/.changeset/resync-model-metadata-2026-10-19.md
new file mode 100644
index 0000000..41c1883
--- /dev/null
+++ b/.changeset/resync-model-metadata-2026-10-19.md
@@ -0,0 +1,5 @@
+---
+"opencode-bridge-for-copilot-chat": patch
+---
+
+Resync supplemental model mirrors with current models.dev metadata.
diff --git a/src/models/metadata.ts b/src/models/metadata.ts
index 2912114..0192d34 100644
--- a/src/models/metadata.ts
+++ b/src/models/metadata.ts
@@ -129,10 +129,7 @@ const SUPPLEMENTAL_MODELS: Readonly<Record<"opencode" | "opencode-go", Readonly<
       family: "glm",
       limit: { context: 204_800, output: 131_072 },
       reasoning: true,
-      // Live-verified 2026-10-09: the Go gateway routes glm-5.1/5.2/5.3 to a
-      // GLM-5.3 backend that only accepts reasoning_effort low|high|max —
-      // a thinking toggle or "none" is rejected ("thinking-only model").
-      reasoning_options: [{ type: "effort", values: ["low", "high", "max"] }],
+      reasoning_options: [{ type: "toggle" }],
       tool_call: true,
       modalities: { input: ["text"] },
       cost: { input: 1.4, output: 4.4, cache_read: 0.26 },
@@ -143,8 +140,7 @@ const SUPPLEMENTAL_MODELS: Readonly<Record<"opencode" | "opencode-go", Readonly<
       family: "glm",
       limit: { context: 204_800, output: 131_072 },
       reasoning: true,
-      // Mirrors the live-verified glm-5.1/5.2/5.3 effort set (see above).
-      reasoning_options: [{ type: "effort", values: ["low", "high", "max"] }],
+      reasoning_options: [{ type: "toggle" }],
       tool_call: true,
       modalities: { input: ["text"] },
       cost: { input: 1, output: 3.2, cache_read: 0.2 },