diff --git a/app/(main)/evaluations/page.tsx b/app/(main)/evaluations/page.tsx
index 28d4c74..15d1522 100644
--- a/app/(main)/evaluations/page.tsx
+++ b/app/(main)/evaluations/page.tsx
@@ -19,7 +19,7 @@ import { FeatureGateModal, LoginModal } from "@/app/components/auth";
 import { Loader, TabNavigation } from "@/app/components/ui";
 import { useToast } from "@/app/hooks/useToast";
 import { DatasetsTab, EvaluationsTab } from "@/app/components/evaluations";
-import { RunMode, Tab } from "@/app/lib/types/evaluation";
+import { Tab } from "@/app/lib/types/evaluation";
 
 const leftPanelWidth = 450;
 
@@ -62,7 +62,6 @@ function SimplifiedEvalContent() {
     },
   );
   const [isEvaluating, setIsEvaluating] = useState(false);
-  const [runMode, setRunMode] = useState<RunMode>("batch");
   const [nameError, setNameError] = useState<string>("");
   const [submitError, setSubmitError] = useState<string>("");
 
@@ -223,14 +222,11 @@ function SimplifiedEvalContent() {
 
     setIsEvaluating(true);
     try {
-      // `run_mode` is only sent when "fast" — omitting it means the backend
-      // defaults to batch, which is the safe behaviour for older clients too.
       const payload: Record<string, unknown> = {
         dataset_id: parseInt(selectedDatasetId),
         experiment_name: experimentName.trim(),
         config_id: selectedConfigId,
         config_version: selectedConfigVersion,
-        run_mode: runMode === "fast" ? "fast" : "batch",
       };
 
       await apiFetch("/api/evaluations", apiKey, {
@@ -243,7 +239,6 @@ function SimplifiedEvalContent() {
       setSelectedDatasetId("");
       setSelectedConfigId("");
       setSelectedConfigVersion(0);
-      setRunMode("batch");
       toast.success(`Evaluation created!`);
       return true;
     } catch (error: unknown) {
@@ -254,12 +249,12 @@ function SimplifiedEvalContent() {
           break;
         case "config_type_unsupported":
           setSubmitError(
-            "Fast mode only supports text-evaluation configs. Pick a text config or switch to Batch.",
+            "This evaluation only supports text-evaluation configs. Pick a text config to continue.",
           );
           break;
         case "dataset_too_large_for_fast":
           setSubmitError(
-            "This dataset is too large for Fast mode (limit is ~10 unique rows / 50 items). Pick a smaller dataset or switch to Batch.",
+            "This dataset is too large for evaluation (limit is 500 items). Pick a smaller dataset.",
           );
           break;
         default:
@@ -349,8 +344,6 @@ function SimplifiedEvalContent() {
               isEvaluating={isEvaluating}
               handleRunEvaluation={handleRunEvaluation}
               setActiveTab={setActiveTab}
-              runMode={runMode}
-              setRunMode={setRunMode}
               nameError={nameError}
               submitError={submitError}
             />
diff --git a/app/components/evaluations/EvaluationsTab.tsx b/app/components/evaluations/EvaluationsTab.tsx
index 418277c..2c8d1a5 100644
--- a/app/components/evaluations/EvaluationsTab.tsx
+++ b/app/components/evaluations/EvaluationsTab.tsx
@@ -3,12 +3,7 @@
 import { useState, useEffect, useCallback } from "react";
 import { apiFetch } from "@/app/lib/apiClient";
 import { Dataset } from "@/app/lib/types/dataset";
-import {
-  AssistantConfig,
-  EvalJob,
-  RunMode,
-  Tab,
-} from "@/app/lib/types/evaluation";
+import { AssistantConfig, EvalJob, Tab } from "@/app/lib/types/evaluation";
 import { useAuth } from "@/app/lib/context/AuthContext";
 import { Modal } from "@/app/components/ui";
 import EvalRunsList from "./EvalRunsList";
@@ -28,8 +23,6 @@ export interface EvaluationsTabProps {
   isEvaluating: boolean;
   handleRunEvaluation: () => Promise<boolean>;
   setActiveTab: (tab: Tab) => void;
-  runMode: RunMode;
-  setRunMode: (mode: RunMode) => void;
   nameError?: string;
   submitError?: string;
 }
@@ -48,8 +41,6 @@ export default function EvaluationsTab({
   isEvaluating,
   handleRunEvaluation,
   setActiveTab,
-  runMode,
-  setRunMode,
   nameError,
   submitError,
 }: EvaluationsTabProps) {
@@ -65,15 +56,14 @@ export default function EvaluationsTab({
   const selectedDataset = storedDatasets.find(
     (d) => d.dataset_id.toString() === selectedDatasetId,
   );
-  const fastModeMismatch =
-    runMode === "fast" && selectedDataset?.eligible_for_fast === false;
+  const datasetTooLarge = selectedDataset?.eligible_for_fast === false;
   const canRun = Boolean(
     experimentName.trim() &&
     selectedDatasetId &&
     selectedConfigId &&
     selectedConfigVersion &&
     !isEvaluating &&
-    !fastModeMismatch,
+    !datasetTooLarge,
   );
 
   const { isAuthenticated } = useAuth();
@@ -158,8 +148,6 @@ export default function EvaluationsTab({
     canRun,
     onRun: handleRun,
     setActiveTab,
-    runMode,
-    setRunMode,
     nameError,
     submitError,
   };
diff --git a/app/components/evaluations/RunEvaluationForm.tsx b/app/components/evaluations/RunEvaluationForm.tsx
index 2236378..9f88e8b 100644
--- a/app/components/evaluations/RunEvaluationForm.tsx
+++ b/app/components/evaluations/RunEvaluationForm.tsx
@@ -1,17 +1,11 @@
 "use client";
 
 import { Dataset } from "@/app/lib/types/dataset";
-import {
-  Button,
-  Field,
-  InfoTooltip,
-  RadioGroup,
-  Select,
-} from "@/app/components/ui";
+import { Button, Field, Select } from "@/app/components/ui";
 import { CheckCircleIcon, PlayIcon } from "@/app/components/icons";
 import ConfigSelector from "@/app/components/ConfigSelector";
 import EvalDatasetDescription from "./EvalDatasetDescription";
-import { RunMode, Tab } from "@/app/lib/types/evaluation";
+import { Tab } from "@/app/lib/types/evaluation";
 import { MAX_NAME_LENGTH } from "@/app/lib/constants";
 
 interface RunEvaluationFormProps {
@@ -28,8 +22,6 @@ interface RunEvaluationFormProps {
   canRun: boolean;
   onRun: () => void;
   setActiveTab: (tab: Tab) => void;
-  runMode: RunMode;
-  setRunMode: (mode: RunMode) => void;
   nameError?: string;
   submitError?: string;
 }
@@ -48,12 +40,10 @@ export default function RunEvaluationForm({
   canRun,
   onRun,
   setActiveTab,
-  runMode,
-  setRunMode,
   nameError,
   submitError,
 }: RunEvaluationFormProps) {
-  const fastEligible = selectedDataset?.eligible_for_fast !== false;
+  const datasetEligible = selectedDataset?.eligible_for_fast !== false;
 
   return (
     <div className="flex-1 overflow-auto p-4 space-y-4">
@@ -140,48 +130,11 @@ export default function RunEvaluationForm({
         </div>
       )}
 
-      <div>
-        <label className="inline-flex items-center text-xs font-medium mb-1.5 text-text-secondary">
-          Run Mode
-          <InfoTooltip
-            text={
-              <>
-                <strong>Batch</strong> — the standard mode. Runs every item in
-                the dataset and produces full metrics.
-                <br />
-                <br />
-                <strong>Fast</strong> — short-loop mode for quick iteration.
-                Only works on small datasets (≤10 unique rows / ≤50 items) and
-                text-evaluation configs.
-              </>
-            }
-          />
-        </label>
-        <RadioGroup<RunMode>
-          ariaLabel="Run mode"
-          className="ml-4"
-          value={runMode}
-          onChange={setRunMode}
-          disabled={isEvaluating}
-          options={[
-            { value: "batch", label: "Batch" },
-            {
-              value: "fast",
-              label: "Fast",
-              title:
-                selectedDataset && !fastEligible
-                  ? "This dataset is too large for Fast mode — pick a smaller one"
-                  : undefined,
-            },
-          ]}
-        />
-        {selectedDataset && !fastEligible && (
-          <p className="text-xs mt-1.5 text-status-error-text">
-            Fast mode isn&apos;t available for this dataset — pick a smaller one
-            or stay on Batch.
-          </p>
-        )}
-      </div>
+      {selectedDataset && !datasetEligible && (
+        <p className="text-xs text-status-error-text">
+          This dataset exceeds the evaluation size limit — pick a smaller one.
+        </p>
+      )}
 
       {submitError && (
         <div className="rounded-lg px-3 py-2 bg-status-error-bg border border-status-error-border">