sentry-junior · Oct 26, 2026
diff --git a/packages/docs/src/content/docs/reference/config-and-env.md b/packages/docs/src/content/docs/reference/config-and-env.md
@@ -28,9 +28,9 @@ related:
 | `JUNIOR_CROSS_ACTOR_MID_RUN_MODE`           | No          | Cross-actor Slack steering policy. Defaults to `follow_up`; see below.                                                                                                      |
 | `AI_MODEL`                                  | No          | Deprecated profile setting. Creates `standard` and remains the fallback for `AI_FAST_MODEL`. Defaults to `openai/gpt-6-luna` with high reasoning.                           |
 | `AI_REASONING_LEVEL`                        | No          | Main-agent reasoning override: `none`, `low`, `medium`, `high`, or `xhigh`. Unset by default. An explicit profile reasoning level takes precedence when routing is enabled. |
-| `AI_FAST_MODEL`                             | No          | Faster model for lightweight tasks and routing/classification passes before the main turn begins. Defaults to `openai/gpt-6-luna`.                                          |
+| `AI_FAST_MODEL`                             | No          | Faster model for lightweight tasks and utility calls. Defaults to `openai/gpt-6-luna`.                                                                                      |
 | `AI_GUARDIAN_MODEL`                         | No          | Model for Guardian action review. Defaults to `openai/gpt-6-luna`.                                                                                                          |
-| `AI_DECISION_MODEL`                         | No          | Decision model for passive reply routing in subscribed threads. Uses the AI SDK decision API. Defaults to `typesafe-ai/jev`.                                                |
+| `AI_DECISION_MODEL`                         | No          | Decision model for the turn router and for passive reply routing in subscribed threads. Uses the AI SDK decision API. Defaults to `typesafe-ai/jev`.                        |
 | `AI_HANDOFF_MODEL`                          | No          | Deprecated profile setting. Creates `handoff`. Defaults to `anthropic/claude-opus-5.5` with high reasoning.                                                                 |
 | `AI_MODEL_PROFILES`                         | No          | Deprecated JSON map of profile names to model IDs for env-only setup. Names must match `^[a-z][a-z0-9_-]*$`.                                                                |
 | `AI_EMBEDDING_MODEL`                        | No          | Embedding model for plugin-owned vector retrieval. Defaults to `openai/text-embedding-3-small`; memory v1 stores fixed 1536-dimensional vectors.                            |

diff --git a/packages/junior-evals/README.md b/packages/junior-evals/README.md
@@ -14,7 +14,7 @@ There are four independently runnable suites:
 - Integration and behavioral cases run the real agent through the agent test fixture.
 - Guardian and router cases use `describeEval()` with their own harness. They call one model boundary and do not run the agent.
 - Behavioral and integration evals use normal core model configuration, including loaded environment overrides. Without overrides, they use the shared core defaults. Neither suite pins models or adds profiles.
-- Router evals use the shared core profiles and configured fast model. Reasoning cases remove fixed profile levels to test the classifier; profile cases retain the defaults.
+- Router evals use the shared core profiles and configured decision model. Reasoning cases remove fixed profile levels to test the classifier; profile cases retain the defaults.
 
 ## Agent Test Fixture
 

diff --git a/packages/junior-evals/src/router-harness.ts b/packages/junior-evals/src/router-harness.ts
@@ -9,7 +9,7 @@ import {
   type DescribeEvalOptions,
   type JsonValue,
 } from "vitest-evals";
-import { completeObject } from "@/chat/pi/client";
+import { completeDecision } from "@/chat/pi/client";
 import {
   DEFAULT_MODEL_PROFILES,
   type ModelProfileConfig,
@@ -48,11 +48,12 @@ async function routeTask(
     : timeoutSignal;
 
   return selectTurnRoute({
-    completeObject: (args) => completeObject({ ...args, signal: routeSignal }),
+    completeDecision: (args) =>
+      completeDecision({ ...args, signal: routeSignal }),
     conversationContext: input.conversationContext,
     currentProfile: input.currentProfile ?? "standard",
-    fastModelId: botConfig.fastModelId,
     messageText: input.messageText,
+    modelId: botConfig.decisionModelId,
     profiles: input.profiles ?? DEFAULT_MODEL_PROFILES,
   });
 }
@@ -106,7 +107,7 @@ export const routerHarness = createHarness<RouterEvalInput, RouterEvalOutput>({
       ],
       usage: {
         provider: "vercel-ai-gateway",
-        model: botConfig.fastModelId,
+        model: botConfig.decisionModelId,
         ...(route.costUsd !== undefined
           ? { metadata: { costUsd: route.costUsd } }
           : {}),

diff --git a/packages/junior/src/chat/agent/index.ts b/packages/junior/src/chat/agent/index.ts
@@ -64,7 +64,7 @@ import {
 } from "@/chat/credentials/context";
 import {
   GEN_AI_PROVIDER_NAME,
-  completeObject,
+  completeDecision,
   completeText,
   getGatewayApiKey,
   resolveGatewayModel,
@@ -657,7 +657,7 @@ async function executeAgentRunInPrivacyContext(
       // unfinished work there or move a new request to another profile.
       const lastProfile = await loadLastModelProfile({ conversationId });
       turnRoute = await selectTurnRoute({
-        completeObject,
+        completeDecision,
         conversationContext: input.conversationContext,
         context: {
           threadId: conversationId,
@@ -670,8 +670,8 @@ async function executeAgentRunInPrivacyContext(
             ? lastProfile
             : botConfig.defaultProfile,
         currentTurnBlocks: routerBlocks,
-        fastModelId: botConfig.fastModelId,
         messageText: userInput,
+        modelId: botConfig.decisionModelId,
         profiles: botConfig.profiles,
       });
       const routedProfileReasoningLevel = profileConfig(

diff --git a/packages/junior/src/chat/config.ts b/packages/junior/src/chat/config.ts
@@ -79,7 +79,7 @@ const DEFAULT_ASSISTANT_LOADING_MESSAGES = [
 ] as const;
 
 export interface BotModelConfig {
-  /** Decision model used by the passive reply router. */
+  /** Decision model used by the turn router and the passive reply router. */
   decisionModelId?: string;
   /** Model used for embeddings. */
   embeddingModelId?: string;