diff --git a/packages/core/src/plugin/provider/lmstudio.ts b/packages/core/src/plugin/provider/lmstudio.ts
index ada8996bb98..3266f302615 100644
--- a/packages/core/src/plugin/provider/lmstudio.ts
+++ b/packages/core/src/plugin/provider/lmstudio.ts
@@ -70,13 +70,11 @@ export function make(origin = "http://127.0.0.1:1234", interval: Duration.Input
               input: ["text", ...(item.capabilities?.vision ? ["image"] : [])],
               output: ["text"],
             }
-            model.limit = {
-              context:
-                item.loaded_instances.length === 0
-                  ? item.max_context_length
-                  : Math.min(...item.loaded_instances.map((instance) => instance.config.context_length)),
-              output: 0,
-            }
+            const context =
+              item.loaded_instances.length === 0
+                ? item.max_context_length
+                : Math.min(...item.loaded_instances.map((instance) => instance.config.context_length))
+            if (context > 0) model.limit.context = context
           })
         }
       })
diff --git a/packages/core/src/plugin/provider/ollama.ts b/packages/core/src/plugin/provider/ollama.ts
index 2b0e03def91..3e6bc8648ad 100644
--- a/packages/core/src/plugin/provider/ollama.ts
+++ b/packages/core/src/plugin/provider/ollama.ts
@@ -96,13 +96,10 @@ export function make(origin = "http://127.0.0.1:11434", interval: Duration.Input
               input: ["text", ...(item.show.capabilities?.includes("vision") ? ["image"] : [])],
               output: ["text"],
             }
-            model.limit = {
-              context:
-                Object.entries(item.show.model_info ?? {}).flatMap(([key, value]) =>
-                  key.endsWith(".context_length") && typeof value === "number" && value > 0 ? [value] : [],
-                )[0] ?? 0,
-              output: 0,
-            }
+            const context = Object.entries(item.show.model_info ?? {}).flatMap(([key, value]) =>
+              key.endsWith(".context_length") && typeof value === "number" && value > 0 ? [value] : [],
+            )[0]
+            if (context !== undefined) model.limit.context = context
           })
         }
       })
diff --git a/packages/core/src/plugin/provider/vllm.ts b/packages/core/src/plugin/provider/vllm.ts
index 0950f4e0b0c..3f723870b14 100644
--- a/packages/core/src/plugin/provider/vllm.ts
+++ b/packages/core/src/plugin/provider/vllm.ts
@@ -55,7 +55,8 @@ export function make(origin = "http://127.0.0.1:8000", interval: Duration.Input
             model.name = item.id
             // Tool calling depends on vLLM server flags and parsers that model discovery does not report.
             model.capabilities = { tools: false, input: ["text"], output: ["text"] }
-            model.limit = { context: item.max_model_len ?? 0, output: 0 }
+            if (typeof item.max_model_len === "number" && item.max_model_len > 0)
+              model.limit.context = item.max_model_len
           })
         }
       })
diff --git a/packages/core/test/plugin/provider-lmstudio.test.ts b/packages/core/test/plugin/provider-lmstudio.test.ts
index f2b4b423c8e..e434b0899b3 100644
--- a/packages/core/test/plugin/provider-lmstudio.test.ts
+++ b/packages/core/test/plugin/provider-lmstudio.test.ts
@@ -71,6 +71,13 @@ describe("LMStudioPlugin", () => {
                   max_context_length: 131_072,
                   capabilities: { vision: false, trained_for_tool_use: false },
                 },
+                {
+                  type: "llm",
+                  key: "unknown-context",
+                  display_name: "Unknown Context",
+                  loaded_instances: [],
+                  max_context_length: 0,
+                },
                 {
                   type: "embedding",
                   key: "nomic-embed",
@@ -104,11 +111,14 @@ describe("LMStudioPlugin", () => {
             family: "gemma4",
             name: "Gemma 4 26B A4B",
             capabilities: { tools: true, input: ["text", "image"], output: ["text"] },
-            limit: { context: 16_384, output: 0 },
+            limit: { context: 16_384, output: 32_000 },
           })
           expect(yield* catalog.model.get(providerID, Model.ID.make("deepseek-r1"))).toMatchObject({
             capabilities: { tools: false, input: ["text"], output: ["text"] },
-            limit: { context: 131_072, output: 0 },
+            limit: { context: 131_072, output: 32_000 },
+          })
+          expect(yield* catalog.model.get(providerID, Model.ID.make("unknown-context"))).toMatchObject({
+            limit: { context: 200_000, output: 32_000 },
           })
           expect(yield* catalog.model.get(providerID, Model.ID.make("nomic-embed"))).toBeUndefined()
         }),
diff --git a/packages/core/test/plugin/provider-ollama.test.ts b/packages/core/test/plugin/provider-ollama.test.ts
index f3f00ebe41e..1db57979df6 100644
--- a/packages/core/test/plugin/provider-ollama.test.ts
+++ b/packages/core/test/plugin/provider-ollama.test.ts
@@ -54,6 +54,7 @@ describe("OllamaPlugin", () => {
                 return Response.json({
                   models: [
                     summary("gemma3:4b", "gemma-digest", "gemma3"),
+                    summary("unknown-context", "unknown-digest"),
                     summary("nomic-embed", "embed-digest"),
                     summary("removed-model", "removed-digest"),
                   ],
@@ -68,6 +69,8 @@ describe("OllamaPlugin", () => {
                       capabilities: ["completion", "tools", "vision"],
                       model_info: { "gemma3.context_length": 131_072 },
                     }
+                  : body.model === "unknown-context"
+                    ? show({ family: "unknown", capabilities: ["completion"], context: 0 })
                   : show({ family: "nomic-bert", capabilities: ["embedding"], context: 8192 }),
               )
             },
@@ -98,7 +101,10 @@ describe("OllamaPlugin", () => {
             name: "gemma3:4b",
             family: "gemma3",
             capabilities: { tools: true, input: ["text", "image"], output: ["text"] },
-            limit: { context: 131_072, output: 0 },
+            limit: { context: 131_072, output: 32_000 },
+          })
+          expect(yield* catalog.model.get(providerID, Model.ID.make("unknown-context"))).toMatchObject({
+            limit: { context: 200_000, output: 32_000 },
           })
           expect(yield* catalog.model.get(providerID, Model.ID.make("nomic-embed"))).toBeUndefined()
           expect(requests).toContainEqual({ method: "GET", path: "/api/tags" })
diff --git a/packages/core/test/plugin/provider-vllm.test.ts b/packages/core/test/plugin/provider-vllm.test.ts
index fb70390360c..3554e45fb4e 100644
--- a/packages/core/test/plugin/provider-vllm.test.ts
+++ b/packages/core/test/plugin/provider-vllm.test.ts
@@ -63,7 +63,11 @@ describe("VLLMPlugin", () => {
               state.models++
               return Response.json({
                 object: "list",
-                data: [remoteModel("Qwen/Qwen3-Coder", 65_536), remoteModel("foreign-model", 4096, "other")],
+                data: [
+                  remoteModel("Qwen/Qwen3-Coder", 65_536),
+                  remoteModel("unknown-limit", 0),
+                  remoteModel("foreign-model", 4096, "other"),
+                ],
               })
             },
           }),
@@ -97,7 +101,10 @@ describe("VLLMPlugin", () => {
             modelID: "Qwen/Qwen3-Coder",
             name: "Qwen/Qwen3-Coder",
             capabilities: { tools: false, input: ["text"], output: ["text"] },
-            limit: { context: 65_536, output: 0 },
+            limit: { context: 65_536, output: 32_000 },
+          })
+          expect(yield* catalog.model.get(providerID, Model.ID.make("unknown-limit"))).toMatchObject({
+            limit: { context: 200_000, output: 32_000 },
           })
           expect(yield* catalog.model.get(providerID, Model.ID.make("foreign-model"))).toBeUndefined()
         }),