diff --git a/packages/core/src/models-dev.ts b/packages/core/src/models-dev.ts
index c2325cb57fda..660041e3cfbc 100644
--- a/packages/core/src/models-dev.ts
+++ b/packages/core/src/models-dev.ts
@@ -195,9 +195,9 @@ const layer = Layer.effect(
     const populate = Effect.gen(function* () {
       const fromDisk = yield* loadFromDisk
       if (fromDisk) return fromDisk
+      if (Flag.OPENCODE_DISABLE_MODELS_FETCH) return {}
       const snapshot = yield* loadSnapshot
       if (snapshot) return snapshot
-      if (Flag.OPENCODE_DISABLE_MODELS_FETCH) return {}
       // Flock is cross-process: concurrent opencode CLIs can race on this cache file.
       const text = yield* Effect.scoped(
         Effect.gen(function* () {
diff --git a/packages/opencode/src/provider/provider.ts b/packages/opencode/src/provider/provider.ts
index c979fc44659c..1e8b342d0fb6 100644
--- a/packages/opencode/src/provider/provider.ts
+++ b/packages/opencode/src/provider/provider.ts
@@ -1619,6 +1619,13 @@ const layer = Layer.effect(
                 (v) => omit(v, ["disabled"]),
               )
             }
+
+            const configLimit = configProvider?.models?.[modelID]?.limit
+            if (configLimit) {
+              if (configLimit.context) model.limit.context = configLimit.context
+              if (configLimit.input) model.limit.input = configLimit.input
+              if (configLimit.output) model.limit.output = configLimit.output
+            }
           }
 
           if (Object.keys(provider.models).length === 0) {
diff --git a/packages/opencode/test/provider/provider.test.ts b/packages/opencode/test/provider/provider.test.ts
index 18ec8f9fbeeb..55793205d69b 100644
--- a/packages/opencode/test/provider/provider.test.ts
+++ b/packages/opencode/test/provider/provider.test.ts
@@ -1214,6 +1214,31 @@ it.instance(
   },
 )
 
+it.instance(
+  "config provider model limit overrides individual fields",
+  Effect.gen(function* () {
+    const providers = yield* list
+    const model = providers[ProviderV2.ID.make("limit-override")].models[ModelV2.ID.make("model")]
+    expect(model.limit.context).toBe(100000)
+    expect(model.limit.input).toBe(80000)
+    expect(model.limit.output).toBe(16000)
+  }),
+  {
+    config: {
+      provider: {
+        "limit-override": {
+          name: "Limit Override Provider",
+          npm: "@ai-sdk/openai-compatible",
+          env: [],
+          models: { model: { name: "Model", tool_call: true, limit: { context: 100000, input: 80000, output: 16000 } } },
+          options: { apiKey: "test" },
+        },
+      },
+    },
+  },
+)
+
+
 it.instance(
   "provider options are deeply merged",
   Effect.gen(function* () {