fix: Simplify configurable model groups (#415)

* fix: Simplify configurable model groups

* cr

apps/open-swe/src/graphs/manager/nodes/classify-message/index.ts

@@ -87,7 +87,7 @@ export async function classifyMessage(
     description: "Respond to the user's message and determine how to route it.",
     schema,
   };
-  const model = await loadModel(config, Task.CLASSIFICATION);
+  const model = await loadModel(config, Task.ROUTER);
   const modelWithTools = model.bindTools([respondAndRouteTool], {
     tool_choice: respondAndRouteTool.name,
     parallel_tool_calls: false,

apps/open-swe/src/graphs/manager/utils/generate-issue-fields.ts

@@ -8,7 +8,7 @@ export async function createIssueFieldsFromMessages(
   messages: BaseMessage[],
   configurable: GraphConfig["configurable"],
 ): Promise<{ title: string; body: string }> {
-  const model = await loadModel({ configurable }, Task.ACTION_GENERATOR);
+  const model = await loadModel({ configurable }, Task.SUMMARIZER);
   const githubIssueTool = {
     name: "create_github_issue",
     description: "Create a new GitHub issue with the given title and body.",

apps/open-swe/src/graphs/planner/nodes/determine-needs-context.ts

@@ -113,7 +113,7 @@ export async function determineNeedsContext(
 ): Promise<Command> {
   const [missingMessages, model] = await Promise.all([
     getMissingMessages(state, config),
-    loadModel(config, Task.CLASSIFICATION),
+    loadModel(config, Task.ROUTER),
   ]);
   if (!missingMessages.length) {
     throw new Error(

apps/open-swe/src/graphs/planner/nodes/generate-message/index.ts

@@ -50,7 +50,7 @@ export async function generateAction(
   state: PlannerGraphState,
   config: GraphConfig,
 ): Promise<PlannerGraphUpdate> {
-  const model = await loadModel(config, Task.ACTION_GENERATOR);
+  const model = await loadModel(config, Task.PROGRAMMER);
   const mcpTools = await getMcpTools(config);
 
   const tools = [

apps/open-swe/src/graphs/planner/nodes/generate-plan/index.ts

@@ -49,7 +49,7 @@ export async function generatePlan(
   state: PlannerGraphState,
   config: GraphConfig,
 ): Promise<PlannerGraphUpdate> {
-  const model = await loadModel(config, Task.PLANNER);
+  const model = await loadModel(config, Task.PROGRAMMER);
   const sessionPlanTool = createSessionPlanToolFields();
   const modelWithTools = model.bindTools([sessionPlanTool], {
     tool_choice: sessionPlanTool.name,

apps/open-swe/src/graphs/planner/nodes/rewrite-plan.ts

@@ -240,7 +240,7 @@ export async function rewritePlan(
     throw new Error("No plan change request found.");
   }
 
-  const model = await loadModel(config, Task.PLANNER);
+  const model = await loadModel(config, Task.PROGRAMMER);
   const tasksToModify = await identifyTasksToModify(state, model);
   const updatedPlanTasks = await updatePlanTasks(state, tasksToModify, model);
 

apps/open-swe/src/graphs/programmer/nodes/generate-message/index.ts

@@ -87,7 +87,7 @@ export async function generateAction(
   state: GraphState,
   config: GraphConfig,
 ): Promise<GraphUpdate> {
-  const model = await loadModel(config, Task.ACTION_GENERATOR);
+  const model = await loadModel(config, Task.PROGRAMMER);
   const mcpTools = await getMcpTools(config);
 
   const tools = [

apps/open-swe/src/graphs/programmer/nodes/open-pr.ts

@@ -88,7 +88,8 @@ export async function openPullRequest(
   }
 
   const openPrTool = createOpenPrToolFields();
-  const model = await loadModel(config, Task.SUMMARIZER);
+  // use the router model since this is a simple task that doesn't need an advanced model
+  const model = await loadModel(config, Task.ROUTER);
   const modelWithTool = model.bindTools([openPrTool], {
     tool_choice: openPrTool.name,
     parallel_tool_calls: false,

apps/open-swe/src/graphs/programmer/nodes/progress-plan-step.ts

@@ -65,7 +65,7 @@ export async function progressPlanStep(
 ): Promise<Command> {
   const markNotCompletedTool = createMarkTaskNotCompletedToolFields();
   const markCompletedTool = createMarkTaskCompletedToolFields();
-  const model = await loadModel(config, Task.PROGRESS_PLAN_CHECKER);
+  const model = await loadModel(config, Task.SUMMARIZER);
   const modelWithTools = model.bindTools(
     [markNotCompletedTool, markCompletedTool],
     {

apps/open-swe/src/graphs/programmer/nodes/update-plan.ts

@@ -117,7 +117,7 @@ export async function updatePlan(
     ...updatePlanToolCall,
   });
 
-  const model = await loadModel(config, Task.PLANNER);
+  const model = await loadModel(config, Task.PROGRAMMER);
   const modelWithTools = model.bindTools([updatePlanTool], {
     tool_choice: updatePlanTool.name,
     parallel_tool_calls: false,

apps/open-swe/src/graphs/reviewer/nodes/final-review.ts

@@ -75,7 +75,7 @@ export async function finalReview(
   const completedTool = createCodeReviewMarkTaskCompletedFields();
   const incompleteTool = createCodeReviewMarkTaskNotCompleteFields();
   const tools = [completedTool, incompleteTool];
-  const model = await loadModel(config, Task.PLANNER);
+  const model = await loadModel(config, Task.PROGRAMMER);
   const modelWithTools = model.bindTools(tools, {
     tool_choice: "any",
     parallel_tool_calls: false,

apps/open-swe/src/graphs/reviewer/nodes/generate-review-actions/index.ts

@@ -73,7 +73,7 @@ export async function generateReviewActions(
   state: ReviewerGraphState,
   config: GraphConfig,
 ): Promise<ReviewerGraphUpdate> {
-  const model = await loadModel(config, Task.ACTION_GENERATOR);
+  const model = await loadModel(config, Task.PROGRAMMER);
   const tools = [
     createSearchTool(state),
     createShellTool(state),

apps/open-swe/src/utils/load-model.ts

@@ -2,39 +2,37 @@ import { initChatModel } from "langchain/chat_models/universal";
 import { GraphConfig } from "@open-swe/shared/open-swe/types";
 
 export enum Task {
-  PLANNER = "planner",
-  PLANNER_CONTEXT = "plannerContext",
-  ACTION_GENERATOR = "actionGenerator",
-  PROGRESS_PLAN_CHECKER = "progressPlanChecker",
+  /**
+   * Used for programmer tasks. This includes: writing code,
+   * generating plans, taking context gathering actions, etc.
+   */
+  PROGRAMMER = "programmer",
+  /**
+   * Used for routing tasks. This includes: initial request
+   * routing to different agents.
+   */
+  ROUTER = "router",
+  /**
+   * Used for summarizing tasks. This includes: summarizing
+   * the conversation history, summarizing actions taken during
+   * a task execution, etc. Should be a slightly advanced model.
+   */
   SUMMARIZER = "summarizer",
-  CLASSIFICATION = "classification",
 }
 
 const TASK_TO_CONFIG_DEFAULTS_MAP = {
-  [Task.PLANNER]: {
+  [Task.PROGRAMMER]: {
     modelName: "anthropic:claude-sonnet-4-0",
     temperature: 0,
   },
-  [Task.PLANNER_CONTEXT]: {
-    modelName: "anthropic:claude-sonnet-4-0",
-    temperature: 0,
-  },
-  [Task.ACTION_GENERATOR]: {
-    modelName: "anthropic:claude-sonnet-4-0",
-    temperature: 0,
-  },
-  [Task.PROGRESS_PLAN_CHECKER]: {
-    modelName: "anthropic:claude-sonnet-4-0",
+  [Task.ROUTER]: {
+    modelName: "anthropic:claude-3-5-haiku-latest",
     temperature: 0,
   },
   [Task.SUMMARIZER]: {
     modelName: "anthropic:claude-sonnet-4-0",
     temperature: 0,
   },
-  [Task.CLASSIFICATION]: {
-    modelName: "anthropic:claude-3-5-haiku-latest",
-    temperature: 0,
-  },
 };
 
 export async function loadModel(config: GraphConfig, task: Task) {

packages/shared/src/open-swe/types.ts

@@ -291,15 +291,16 @@ export const GraphConfigurationMetadata: {
       description: "Maximum number of review actions during planning",
     },
   },
-  plannerModelName: {
+  programmerModelName: {
     x_open_swe_ui_config: {
       type: "select",
       default: "anthropic:claude-sonnet-4-0",
-      description: "The model to use for planning",
+      description:
+        "The model to use for programming/other advanced technical tasks",
       options: MODEL_OPTIONS_NO_THINKING,
     },
   },
-  plannerTemperature: {
+  programmerTemperature: {
     x_open_swe_ui_config: {
       type: "slider",
       default: 0,
@@ -309,51 +310,15 @@ export const GraphConfigurationMetadata: {
       description: "Controls randomness (0 = deterministic, 2 = creative)",
     },
   },
-  plannerContextModelName: {
+  routerModelName: {
     x_open_swe_ui_config: {
       type: "select",
-      default: "anthropic:claude-sonnet-4-0",
-      description: "The model to use for planning",
+      default: "anthropic:claude-3-5-haiku-latest",
+      description: "The model to use for routing tasks",
       options: MODEL_OPTIONS,
     },
   },
-  plannerContextTemperature: {
-    x_open_swe_ui_config: {
-      type: "slider",
-      default: 0,
-      min: 0,
-      max: 2,
-      step: 0.1,
-      description: "Controls randomness (0 = deterministic, 2 = creative)",
-    },
-  },
-  actionGeneratorModelName: {
-    x_open_swe_ui_config: {
-      type: "select",
-      default: "anthropic:claude-sonnet-4-0",
-      description: "The model to use for action generation",
-      options: MODEL_OPTIONS,
-    },
-  },
-  actionGeneratorTemperature: {
-    x_open_swe_ui_config: {
-      type: "slider",
-      default: 0,
-      min: 0,
-      max: 2,
-      step: 0.1,
-      description: "Controls randomness (0 = deterministic, 2 = creative)",
-    },
-  },
-  progressPlanCheckerModelName: {
-    x_open_swe_ui_config: {
-      type: "select",
-      default: "anthropic:claude-sonnet-4-0",
-      description: "The model to use for progress plan checking",
-      options: MODEL_OPTIONS_NO_THINKING,
-    },
-  },
-  progressPlanCheckerTemperature: {
+  routerTemperature: {
     x_open_swe_ui_config: {
       type: "slider",
       default: 0,
@@ -367,7 +332,8 @@ export const GraphConfigurationMetadata: {
     x_open_swe_ui_config: {
       type: "select",
       default: "anthropic:claude-sonnet-4-0",
-      description: "The model to use for summarizing the conversation history",
+      description:
+        "The model to use for summarizing the conversation history, or extracting key context from large inputs",
       options: MODEL_OPTIONS_NO_THINKING,
     },
   },
@@ -381,24 +347,6 @@ export const GraphConfigurationMetadata: {
       description: "Controls randomness (0 = deterministic, 2 = creative)",
     },
   },
-  classificationModelName: {
-    x_open_swe_ui_config: {
-      type: "select",
-      default: "anthropic:claude-3-5-haiku-latest",
-      description: "The model to use for classifying the user's message",
-      options: MODEL_OPTIONS_NO_THINKING,
-    },
-  },
-  classificationTemperature: {
-    x_open_swe_ui_config: {
-      type: "slider",
-      default: 0,
-      min: 0,
-      max: 2,
-      step: 0.1,
-      description: "Controls randomness (0 = deterministic, 2 = creative)",
-    },
-  },
   maxTokens: {
     x_open_swe_ui_config: {
       type: "number",
@@ -456,200 +404,124 @@ export const GraphConfiguration = z.object({
    * Total messages = maxContextActions * 2 + 1
    * @default 75
    */
-  maxContextActions: z
-    .number()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.maxContextActions),
+  maxContextActions: withLangGraph(z.number().optional(), {
+    metadata: GraphConfigurationMetadata.maxContextActions,
+  }),
   /**
    * The maximum number of context gathering actions to take during review.
    * Each action consists of 2 messages (request & result), plus 1 human message.
    * Total messages = maxReviewActions * 2 + 1
    * @default 30
    */
-  maxReviewActions: z
-    .number()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.maxReviewActions),
+  maxReviewActions: withLangGraph(z.number().optional(), {
+    metadata: GraphConfigurationMetadata.maxReviewActions,
+  }),
 
   /**
-   * The model ID to use for the planning step.
-   * This includes initial planning, and rewriting.
+   * The model ID to use for programming/other advanced technical tasks.
    * @default "anthropic:claude-sonnet-4-0"
    */
-  plannerModelName: z
-    .string()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.plannerModelName),
+  programmerModelName: withLangGraph(z.string().optional(), {
+    metadata: GraphConfigurationMetadata.programmerModelName,
+  }),
   /**
-   * The temperature to use for the planning step.
-   * This includes initial planning, and rewriting.
-   * If selecting a reasoning model, this will be ignored.
+   * The temperature to use for programming/other advanced technical tasks.
    * @default 0
    */
-  plannerTemperature: z
-    .number()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.plannerTemperature),
+  programmerTemperature: withLangGraph(z.number().optional(), {
+    metadata: GraphConfigurationMetadata.programmerTemperature,
+  }),
   /**
-   * The model ID to use for the planning step.
-   * This includes initial planning, and rewriting.
-   * @default "anthropic:claude-sonnet-4-0"
-   */
-  plannerContextModelName: z
-    .string()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.plannerContextModelName),
-  /**
-   * The temperature to use for the planning step.
-   * This includes initial planning, and rewriting.
-   * If selecting a reasoning model, this will be ignored.
-   * @default 0
-   */
-  plannerContextTemperature: z
-    .number()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.plannerContextTemperature),
-
-  /**
-   * The model ID to use for action generation.
-   * @default "anthropic:claude-sonnet-4-0"
-   */
-  actionGeneratorModelName: z
-    .string()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.actionGeneratorModelName),
-  /**
-   * The temperature to use for action generation.
-   * If selecting a reasoning model, this will be ignored.
-   * @default 0
-   */
-  actionGeneratorTemperature: z
-    .number()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.actionGeneratorTemperature),
-
-  /**
-   * The model ID to use for progress plan checking.
-   * @default "anthropic:claude-sonnet-4-0"
-   */
-  progressPlanCheckerModelName: z
-    .string()
-    .optional()
-    .langgraph.metadata(
-      GraphConfigurationMetadata.progressPlanCheckerModelName,
-    ),
-  /**
-   * The temperature to use for progress plan checking.
-   * If selecting a reasoning model, this will be ignored.
-   * @default 0
-   */
-  progressPlanCheckerTemperature: z
-    .number()
-    .optional()
-    .langgraph.metadata(
-      GraphConfigurationMetadata.progressPlanCheckerTemperature,
-    ),
-
-  /**
-   * The model ID to use for summarizing the conversation history.
-   * @default "anthropic:claude-sonnet-4-0"
-   */
-  summarizerModelName: z
-    .string()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.summarizerModelName),
-  /**
-   * The temperature to use for summarizing the conversation history.
-   * If selecting a reasoning model, this will be ignored.
-   * @default 0
-   */
-  summarizerTemperature: z
-    .number()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.summarizerTemperature),
-
-  /**
-   * The model ID to use for classifying the user's message.
+   * The model ID to use for routing tasks.
    * @default "anthropic:claude-3-5-haiku-latest"
    */
-  classificationModelName: z
-    .string()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.classificationModelName),
+  routerModelName: withLangGraph(z.string().optional(), {
+    metadata: GraphConfigurationMetadata.routerModelName,
+  }),
   /**
-   * The temperature to use for classifying the user's message.
-   * If selecting a reasoning model, this will be ignored.
+   * The temperature to use for routing tasks.
    * @default 0
    */
-  classificationTemperature: z
-    .number()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata.classificationTemperature),
+  routerTemperature: withLangGraph(z.number().optional(), {
+    metadata: GraphConfigurationMetadata.routerTemperature,
+  }),
 
+  /**
+   * The model ID to use for summarizing the conversation history, or extracting key context from large inputs.
+   * @default "anthropic:claude-sonnet-4-0"
+   */
+  summarizerModelName: withLangGraph(z.string().optional(), {
+    metadata: GraphConfigurationMetadata.summarizerModelName,
+  }),
+  /**
+   * The temperature to use for summarizing the conversation history, or extracting key context from large inputs.
+   * @default 0
+   */
+  summarizerTemperature: withLangGraph(z.number().optional(), {
+    metadata: GraphConfigurationMetadata.actionGeneratorTemperature,
+  }),
   /**
    * The maximum number of tokens to generate in an individual generation.
    * @default 10_000
    */
-  maxTokens: z
-    .number()
-    .optional()
-    .default(() => 10_000)
-    .langgraph.metadata(GraphConfigurationMetadata.maxTokens),
+  maxTokens: withLangGraph(z.number(), {
+    metadata: GraphConfigurationMetadata.maxTokens,
+    reducer: {
+      schema: z.number(),
+      fn: (_state, update) => update,
+    },
+    default: () => 10_000,
+  }),
   /**
    * The user's GitHub access token. To be used in requests to get information about the user.
    */
-  [GITHUB_TOKEN_COOKIE]: z
-    .string()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata[GITHUB_TOKEN_COOKIE]),
+  [GITHUB_TOKEN_COOKIE]: withLangGraph(z.string().optional(), {
+    metadata: GraphConfigurationMetadata[GITHUB_TOKEN_COOKIE],
+  }),
   /**
    * The installation token from the GitHub app. This token allows us to take actions
    * on the repos the user has granted us access to, but on behalf of the app, not the user.
    */
-  [GITHUB_INSTALLATION_TOKEN_COOKIE]: z
-    .string()
-    .optional()
-    .langgraph.metadata(
-      GraphConfigurationMetadata[GITHUB_INSTALLATION_TOKEN_COOKIE],
-    ),
+  [GITHUB_INSTALLATION_TOKEN_COOKIE]: withLangGraph(z.string().optional(), {
+    metadata: GraphConfigurationMetadata[GITHUB_INSTALLATION_TOKEN_COOKIE],
+  }),
   /**
    * The user's GitHub ID. Required when creating runs triggered by a bot (e.g. GitHub issue)
    */
-  [GITHUB_USER_ID_HEADER]: z
-    .string()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata[GITHUB_USER_ID_HEADER]),
+  [GITHUB_USER_ID_HEADER]: withLangGraph(z.string().optional(), {
+    metadata: GraphConfigurationMetadata[GITHUB_USER_ID_HEADER],
+  }),
   /**
    * The user's GitHub login. Required when creating runs triggered by a bot (e.g. GitHub issue)
    */
-  [GITHUB_USER_LOGIN_HEADER]: z
-    .string()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata[GITHUB_USER_LOGIN_HEADER]),
+  [GITHUB_USER_LOGIN_HEADER]: withLangGraph(z.string().optional(), {
+    metadata: GraphConfigurationMetadata[GITHUB_USER_LOGIN_HEADER],
+  }),
   /**
    * The installation name of the GitHub app. Required when creating runs triggered by a bot (e.g. GitHub issue)
    */
-  [GITHUB_INSTALLATION_NAME]: z
-    .string()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata[GITHUB_INSTALLATION_NAME]),
+  [GITHUB_INSTALLATION_NAME]: withLangGraph(z.string().optional(), {
+    metadata: GraphConfigurationMetadata[GITHUB_INSTALLATION_NAME],
+  }),
   /**
    * GitHub Personal Access Token. Used for simpler authentication in environments like evals
    * where GitHub App installation tokens are not available or needed.
    */
-  [GITHUB_PAT]: z
-    .string()
-    .optional()
-    .langgraph.metadata(GraphConfigurationMetadata[GITHUB_PAT]),
+  [GITHUB_PAT]: withLangGraph(z.string().optional(), {
+    metadata: GraphConfigurationMetadata[GITHUB_PAT],
+  }),
   /**
    * Custom MCP servers configuration as JSON string. Merges with default servers.
    * @default Default LangGraph docs MCP server
    */
-  mcpServers: z
-    .string()
-    .optional()
-    .default(JSON.stringify(DEFAULT_MCP_SERVERS))
-    .langgraph.metadata(GraphConfigurationMetadata.mcpServers),
+  mcpServers: withLangGraph(z.string(), {
+    metadata: GraphConfigurationMetadata.mcpServers,
+    default: () => JSON.stringify(DEFAULT_MCP_SERVERS),
+    reducer: {
+      schema: z.string(),
+      fn: (_state, update) => update,
+    },
+  }),
 });
 
 export type GraphConfig = LangGraphRunnableConfig<