diff --git a/actions/setup/js/effective_tokens.cjs b/actions/setup/js/effective_tokens.cjs
index 2f3f8cd6cd0..9c890d8f79c 100644
--- a/actions/setup/js/effective_tokens.cjs
+++ b/actions/setup/js/effective_tokens.cjs
@@ -212,6 +212,111 @@ function formatET(n) {
   return `${(n / 1_000_000).toFixed(1).replace(/\.0$/, "")}M`;
 }
 
+/**
+ * Build a deterministic 5-character model identifier for footer rendering.
+ * Uses well-known shortcuts for popular model families and a deterministic fallback.
+ *
+ * Examples:
+ * - claude-sonnet-4.6 -> son46
+ * - gpt-5.5 -> gpt55
+ * - claude-opus-4-7 -> opu47
+ *
+ * @param {string|undefined|null} modelName
+ * @returns {string}
+ */
+function reduceModelNameToIdentifier(modelName) {
+  const normalized = String(modelName || "")
+    .trim()
+    .toLowerCase();
+  if (!normalized) return "";
+
+  const VERSION_SUFFIX_PATTERN = "[-_\\s]*([0-9]+)(?:[._-]+([0-9]+))?";
+  const FALLBACK_LETTER_LENGTH = 3;
+  const FALLBACK_DIGIT_LENGTH = 2;
+  const FALLBACK_PADDING_CHAR = "x";
+
+  /** @type {Array<{ familyPattern: RegExp, versionPattern: RegExp, prefix: string }>} */
+  const shortcuts = [
+    { familyPattern: /sonnet/, versionPattern: new RegExp(`sonnet${VERSION_SUFFIX_PATTERN}`), prefix: "son" },
+    { familyPattern: /opus/, versionPattern: new RegExp(`opus${VERSION_SUFFIX_PATTERN}`), prefix: "opu" },
+    { familyPattern: /haiku/, versionPattern: new RegExp(`haiku${VERSION_SUFFIX_PATTERN}`), prefix: "hai" },
+    { familyPattern: /gpt/, versionPattern: new RegExp(`gpt${VERSION_SUFFIX_PATTERN}`), prefix: "gpt" },
+    { familyPattern: /gemini/, versionPattern: new RegExp(`gemini${VERSION_SUFFIX_PATTERN}`), prefix: "gem" },
+  ];
+
+  for (const { familyPattern, versionPattern, prefix } of shortcuts) {
+    if (!familyPattern.test(normalized)) continue;
+    const version = extractModelVersionDigits(normalized, versionPattern);
+    return `${prefix}${version}`;
+  }
+
+  return buildFallbackModelIdentifier(normalized, FALLBACK_LETTER_LENGTH, FALLBACK_DIGIT_LENGTH, FALLBACK_PADDING_CHAR);
+}
+
+/**
+ * @param {string} normalizedModelName
+ * @param {RegExp} familyVersionPattern
+ * @returns {string}
+ */
+function extractModelVersionDigits(normalizedModelName, familyVersionPattern) {
+  const familyMatch = normalizedModelName.match(familyVersionPattern);
+  if (familyMatch) {
+    return normalizeVersionDigits(familyMatch[1], familyMatch[2]);
+  }
+
+  const firstNumericMatch = normalizedModelName.match(/([0-9]+)(?:[._-]+([0-9]+))?/);
+  if (firstNumericMatch) {
+    return normalizeVersionDigits(firstNumericMatch[1], firstNumericMatch[2]);
+  }
+
+  return "00";
+}
+
+/**
+ * @param {string|undefined} major
+ * @param {string|undefined} minor
+ * @returns {string}
+ */
+function normalizeVersionDigits(major, minor) {
+  const majorDigit = getFirstDigit(major);
+  // Treat any 3+ digit minor segment as a build/date-like stamp (e.g. 100, 20250514),
+  // not a semantic minor version, so identifiers stay stable (gpt-5-2025-08-07 -> gpt50).
+  const minorIsDateLike = minor && /^\d{3,}$/.test(minor);
+  const minorDigit = getFirstDigit(minor, Boolean(minorIsDateLike));
+  return `${majorDigit}${minorDigit}`;
+}
+
+/**
+ * @param {string|undefined} value
+ * @param {boolean} [treatAsMissing=false]
+ * @returns {string}
+ */
+function getFirstDigit(value, treatAsMissing = false) {
+  if (!value || treatAsMissing) return "0";
+  const digitMatch = value.match(/\d/);
+  return digitMatch ? digitMatch[0] : "0";
+}
+
+/**
+ * @param {string} normalizedModelName
+ * @param {number} fallbackLetterLength
+ * @param {number} fallbackDigitLength
+ * @param {string} fallbackPaddingChar
+ * @returns {string}
+ */
+function buildFallbackModelIdentifier(normalizedModelName, fallbackLetterLength, fallbackDigitLength, fallbackPaddingChar) {
+  const compact = normalizedModelName.replace(/[^a-z0-9]+/g, "");
+  if (!compact) return "";
+
+  // Pad with "x" to keep a fixed family slot for short/unknown model names.
+  const letterPart = compact.replace(/[0-9]/g, "").slice(0, fallbackLetterLength).padEnd(fallbackLetterLength, fallbackPaddingChar);
+  const digitPart = compact
+    .replace(/[^0-9]/g, "")
+    .slice(0, fallbackDigitLength)
+    .padEnd(fallbackDigitLength, "0");
+  return `${letterPart}${digitPart}`.slice(0, 5);
+}
+
 /**
  * Resets the cached multipliers (for testing purposes).
  * @internal
@@ -231,7 +336,9 @@ function getEffectiveTokensSuffix() {
   const parsed = parseInt(raw, 10);
 
   if (!isNaN(parsed) && parsed > 0) {
-    return ` · ● ${formatET(parsed)}`;
+    const reducedModel = reduceModelNameToIdentifier(process.env.GH_AW_ENGINE_MODEL);
+    const modelPrefix = reducedModel ? `${reducedModel} ` : "";
+    return ` · ● ${modelPrefix}${formatET(parsed)}`;
   }
   return "";
 }
@@ -336,6 +443,7 @@ module.exports = {
   computeBaseWeightedTokens,
   computeEffectiveTokens,
   formatET,
+  reduceModelNameToIdentifier,
   getEffectiveTokensSuffix,
   AGENT_USAGE_PATH,
   readAgentUsage,
diff --git a/actions/setup/js/effective_tokens.test.cjs b/actions/setup/js/effective_tokens.test.cjs
index 2e45847fdfa..776044918e9 100644
--- a/actions/setup/js/effective_tokens.test.cjs
+++ b/actions/setup/js/effective_tokens.test.cjs
@@ -1,7 +1,17 @@
 // @ts-check
 /// <reference types="@actions/github-script" />
 
-const { defaultTokenClassWeights, getTokenClassWeights, getModelMultiplier, computeBaseWeightedTokens, computeEffectiveTokens, formatET, _resetCache } = require("./effective_tokens.cjs");
+const {
+  defaultTokenClassWeights,
+  getTokenClassWeights,
+  getModelMultiplier,
+  computeBaseWeightedTokens,
+  computeEffectiveTokens,
+  formatET,
+  reduceModelNameToIdentifier,
+  getEffectiveTokensSuffix,
+  _resetCache,
+} = require("./effective_tokens.cjs");
 
 // Model multipliers JSON used in tests (matches pkg/cli/data/model_multipliers.json)
 const TEST_MULTIPLIERS_JSON = JSON.stringify({
@@ -329,6 +339,69 @@ describe("effective_tokens", () => {
       expect(formatET(999)).toBe("999");
     });
 
+    describe("reduceModelNameToIdentifier", () => {
+      test("returns empty string for null input", () => {
+        expect(reduceModelNameToIdentifier(null)).toBe("");
+      });
+
+      test("returns empty string for undefined input", () => {
+        expect(reduceModelNameToIdentifier(undefined)).toBe("");
+      });
+
+      test("returns empty string for empty string input", () => {
+        expect(reduceModelNameToIdentifier("")).toBe("");
+      });
+
+      test("uses well-known sonnet shortcut", () => {
+        expect(reduceModelNameToIdentifier("claude-sonnet-4.6")).toBe("son46");
+      });
+
+      test("uses well-known gpt shortcut", () => {
+        expect(reduceModelNameToIdentifier("gpt-5.5")).toBe("gpt55");
+      });
+
+      test("uses well-known opus shortcut", () => {
+        expect(reduceModelNameToIdentifier("claude-opus-4-7")).toBe("opu47");
+      });
+
+      test("uses well-known haiku shortcut", () => {
+        expect(reduceModelNameToIdentifier("claude-haiku-4.5")).toBe("hai45");
+      });
+
+      test("uses well-known gemini shortcut", () => {
+        expect(reduceModelNameToIdentifier("gemini-2.5-pro")).toBe("gem25");
+      });
+
+      test("handles date-like suffixes deterministically", () => {
+        expect(reduceModelNameToIdentifier("gpt-5-2025-08-07")).toBe("gpt50");
+        expect(reduceModelNameToIdentifier("claude-sonnet-4-20250514")).toBe("son40");
+        expect(reduceModelNameToIdentifier("gpt-4-100")).toBe("gpt40");
+      });
+
+      test("returns deterministic 5-character fallback for unknown models", () => {
+        expect(reduceModelNameToIdentifier("my-custom-engine-v2")).toBe("myc20");
+      });
+    });
+
+    describe("getEffectiveTokensSuffix", () => {
+      afterEach(() => {
+        delete process.env.GH_AW_EFFECTIVE_TOKENS;
+        delete process.env.GH_AW_ENGINE_MODEL;
+      });
+
+      test("prepends reduced model identifier when model is available", () => {
+        process.env.GH_AW_EFFECTIVE_TOKENS = "12500";
+        process.env.GH_AW_ENGINE_MODEL = "claude-sonnet-4.6";
+        expect(getEffectiveTokensSuffix()).toBe(" · ● son46 12.5K");
+      });
+
+      test("falls back to token-only suffix when model is unavailable", () => {
+        process.env.GH_AW_EFFECTIVE_TOKENS = "12500";
+        delete process.env.GH_AW_ENGINE_MODEL;
+        expect(getEffectiveTokensSuffix()).toBe(" · ● 12.5K");
+      });
+    });
+
     test("formats values in the thousands as K", () => {
       expect(formatET(1000)).toBe("1K");
       expect(formatET(1200)).toBe("1.2K");
diff --git a/actions/setup/js/messages.test.cjs b/actions/setup/js/messages.test.cjs
index 5e096f9ce88..5eff7d8f4c3 100644
--- a/actions/setup/js/messages.test.cjs
+++ b/actions/setup/js/messages.test.cjs
@@ -366,6 +366,20 @@ describe("messages.cjs", () => {
       expect(result).toBe("> Generated by [Test Workflow](https://github.com/test/repo/actions/runs/123) · ● 12.5K");
     });
 
+    it("should prepend reduced model identifier to effective tokens when engine model is set", async () => {
+      process.env.GH_AW_EFFECTIVE_TOKENS = "12500";
+      process.env.GH_AW_ENGINE_MODEL = "claude-sonnet-4.6";
+
+      const { getFooterMessage } = await import("./messages.cjs");
+
+      const result = getFooterMessage({
+        workflowName: "Test Workflow",
+        runUrl: "https://github.com/test/repo/actions/runs/123",
+      });
+
+      expect(result).toBe("> Generated by [Test Workflow](https://github.com/test/repo/actions/runs/123) · ● son46 12.5K");
+    });
+
     it("should include effective tokens from env var before history link", async () => {
       process.env.GH_AW_EFFECTIVE_TOKENS = "5000";
       const historyUrl = "https://github.com/search?q=repo:test/repo+is:issue&type=issues";
@@ -425,6 +439,23 @@ describe("messages.cjs", () => {
 
       expect(result).toBe("> Custom: [Test Workflow](https://github.com/test/repo/actions/runs/123) · ● 5K");
     });
+
+    it("should include reduced model identifier in effective_tokens_suffix for custom templates", async () => {
+      process.env.GH_AW_EFFECTIVE_TOKENS = "5000";
+      process.env.GH_AW_ENGINE_MODEL = "gpt-5.5";
+      process.env.GH_AW_SAFE_OUTPUT_MESSAGES = JSON.stringify({
+        footer: "> Custom: [{workflow_name}]({run_url}){effective_tokens_suffix}",
+      });
+
+      const { getFooterMessage } = await import("./messages.cjs");
+
+      const result = getFooterMessage({
+        workflowName: "Test Workflow",
+        runUrl: "https://github.com/test/repo/actions/runs/123",
+      });
+
+      expect(result).toBe("> Custom: [Test Workflow](https://github.com/test/repo/actions/runs/123) · ● gpt55 5K");
+    });
   });
 
   describe("getFooterInstallMessage", () => {
diff --git a/actions/setup/js/messages_footer.cjs b/actions/setup/js/messages_footer.cjs
index 92a58e31bbb..c5635dd5044 100644
--- a/actions/setup/js/messages_footer.cjs
+++ b/actions/setup/js/messages_footer.cjs
@@ -12,7 +12,7 @@ const { getMessages, renderTemplate, renderTemplateFromFile, toSnakeCase, getPro
 const { getMissingInfoSections } = require("./missing_messages_helper.cjs");
 const { getBlockedDomains, generateBlockedDomainsSection } = require("./firewall_blocked_domains.cjs");
 const { getDifcFilteredEvents, generateDifcFilteredSection } = require("./gateway_difc_filtered.cjs");
-const { formatET } = require("./effective_tokens.cjs");
+const { formatET, reduceModelNameToIdentifier } = require("./effective_tokens.cjs");
 const { getDetectionWarningMessage } = require("./messages_run_status.cjs");
 
 /**
@@ -37,18 +37,29 @@ function getDetectionCautionAlert(workflowName, runUrl) {
  * both the raw count, compact formatted string, and a pre-formatted suffix.
  * Returns undefined/empty for all fields when the variable is absent or the parsed value
  * is not a positive integer.
+ * @param {string} modelName
  * @returns {{ effectiveTokens: number|undefined, effectiveTokensFormatted: string|undefined, effectiveTokensSuffix: string }}
  */
-function getEffectiveTokensFromEnv() {
+function getEffectiveTokensFromEnv(modelName) {
   const raw = process.env.GH_AW_EFFECTIVE_TOKENS;
   const parsed = raw ? parseInt(raw, 10) : NaN;
   if (!isNaN(parsed) && parsed > 0) {
+    const modelPrefix = buildModelPrefix(modelName);
     const effectiveTokensFormatted = formatET(parsed);
-    return { effectiveTokens: parsed, effectiveTokensFormatted, effectiveTokensSuffix: ` · ● ${effectiveTokensFormatted}` };
+    return { effectiveTokens: parsed, effectiveTokensFormatted, effectiveTokensSuffix: ` · ● ${modelPrefix}${effectiveTokensFormatted}` };
   }
   return { effectiveTokens: undefined, effectiveTokensFormatted: undefined, effectiveTokensSuffix: "" };
 }
 
+/**
+ * @param {string} modelName
+ * @returns {string}
+ */
+function buildModelPrefix(modelName) {
+  const reducedModel = reduceModelNameToIdentifier(modelName);
+  return reducedModel ? `${reducedModel} ` : "";
+}
+
 /**
  * @typedef {Object} FooterContext
  * @property {string} workflowName - Name of the workflow
@@ -60,6 +71,7 @@ function getEffectiveTokensFromEnv() {
  * @property {string} [historyUrl] - GitHub search URL for items created by this workflow (for the history link)
  * @property {string} [historyLink] - Pre-formatted markdown history link (e.g. " · [◷](url)"), or "" if unavailable
  * @property {number} [effectiveTokens] - Total effective token count for the run (shown as ● N when > 0, in compact format)
+ * @property {string} [model] - Model name used for the run, used to build a compact model identifier in ET suffixes
  * @property {string} [emoji] - Optional emoji representing the workflow (from frontmatter)
  */
 
@@ -74,7 +86,8 @@ function getFooterMessage(ctx) {
   // Use effectiveTokens from context if provided, otherwise fall back to env var.
   // This ensures callers that don't pass effectiveTokens (e.g. update_activation_comment.cjs)
   // still get the effective token count in the footer when GH_AW_EFFECTIVE_TOKENS is set.
-  const { effectiveTokens: envEffectiveTokens } = getEffectiveTokensFromEnv();
+  const resolvedModelName = ctx.model || process.env.GH_AW_ENGINE_MODEL || "";
+  const { effectiveTokens: envEffectiveTokens, effectiveTokensFormatted: envEffectiveTokensFormatted, effectiveTokensSuffix: envEffectiveTokensSuffix } = getEffectiveTokensFromEnv(resolvedModelName);
   const effectiveTokens = ctx.effectiveTokens ?? envEffectiveTokens;
 
   // Pre-compute history_link as a ready-to-use markdown suffix (empty string when unavailable)
@@ -84,9 +97,19 @@ function getFooterMessage(ctx) {
   const agenticWorkflowUrl = ctx.agenticWorkflowUrl || (ctx.runUrl ? `${ctx.runUrl}/agentic_workflow` : "");
 
   // Pre-compute effective_tokens_formatted and effective_tokens_suffix for use in custom templates
-  const effectiveTokensFormatted = effectiveTokens ? formatET(effectiveTokens) : undefined;
+  const hasExplicitContextEffectiveTokens = ctx.effectiveTokens !== undefined && ctx.effectiveTokens !== null;
+  let effectiveTokensFormatted = envEffectiveTokensFormatted;
   // effective_tokens_suffix is always a string: either " · ● 1.2K" or "" (for safe use in templates)
-  const effectiveTokensSuffix = effectiveTokensFormatted ? ` · ● ${effectiveTokensFormatted}` : "";
+  let effectiveTokensSuffix = envEffectiveTokensSuffix;
+  if (hasExplicitContextEffectiveTokens) {
+    effectiveTokensFormatted = effectiveTokens ? formatET(effectiveTokens) : undefined;
+    if (effectiveTokensFormatted) {
+      const modelPrefix = buildModelPrefix(resolvedModelName);
+      effectiveTokensSuffix = ` · ● ${modelPrefix}${effectiveTokensFormatted}`;
+    } else {
+      effectiveTokensSuffix = "";
+    }
+  }
 
   // Create context with both camelCase and snake_case keys, including computed history_link and agentic_workflow_url
   const templateContext = toSnakeCase({ ...ctx, effectiveTokens, historyLink, agenticWorkflowUrl, effectiveTokensFormatted, effectiveTokensSuffix });
@@ -104,7 +127,7 @@ function getFooterMessage(ctx) {
   }
   // Append effective tokens with ● symbol when available (compact format, no "ET" label)
   if (effectiveTokens) {
-    defaultFooter += ` · ● ${formatET(effectiveTokens)}`;
+    defaultFooter += effectiveTokensSuffix;
   }
   // Append history link when available
   if (ctx.historyUrl) {
@@ -157,7 +180,8 @@ function getFooterWorkflowRecompileMessage(ctx) {
   const agenticWorkflowUrl = ctx.agenticWorkflowUrl || (ctx.runUrl ? `${ctx.runUrl}/agentic_workflow` : "");
 
   // Read effective tokens from environment variable if available
-  const { effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix } = getEffectiveTokensFromEnv();
+  const modelName = process.env.GH_AW_ENGINE_MODEL || "";
+  const { effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix } = getEffectiveTokensFromEnv(modelName);
 
   // Create context with both camelCase and snake_case keys
   const templateContext = toSnakeCase({ ...ctx, agenticWorkflowUrl, effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix });
@@ -183,7 +207,8 @@ function getFooterWorkflowRecompileCommentMessage(ctx) {
   const agenticWorkflowUrl = ctx.agenticWorkflowUrl || (ctx.runUrl ? `${ctx.runUrl}/agentic_workflow` : "");
 
   // Read effective tokens from environment variable if available
-  const { effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix } = getEffectiveTokensFromEnv();
+  const modelName = process.env.GH_AW_ENGINE_MODEL || "";
+  const { effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix } = getEffectiveTokensFromEnv(modelName);
 
   // Create context with both camelCase and snake_case keys
   const templateContext = toSnakeCase({ ...ctx, agenticWorkflowUrl, effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix });
@@ -222,7 +247,8 @@ function getFooterAgentFailureIssueMessage(ctx) {
   const agenticWorkflowUrl = ctx.agenticWorkflowUrl || (ctx.runUrl ? `${ctx.runUrl}/agentic_workflow` : "");
 
   // Read effective tokens from environment variable if available
-  const { effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix } = getEffectiveTokensFromEnv();
+  const modelName = process.env.GH_AW_ENGINE_MODEL || "";
+  const { effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix } = getEffectiveTokensFromEnv(modelName);
 
   // Create context with both camelCase and snake_case keys, including computed history_link and agentic_workflow_url
   const templateContext = toSnakeCase({ ...ctx, historyLink, agenticWorkflowUrl, effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix });
@@ -263,7 +289,8 @@ function getFooterAgentFailureCommentMessage(ctx) {
   const agenticWorkflowUrl = ctx.agenticWorkflowUrl || (ctx.runUrl ? `${ctx.runUrl}/agentic_workflow` : "");
 
   // Read effective tokens from environment variable if available
-  const { effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix } = getEffectiveTokensFromEnv();
+  const modelName = process.env.GH_AW_ENGINE_MODEL || "";
+  const { effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix } = getEffectiveTokensFromEnv(modelName);
 
   // Create context with both camelCase and snake_case keys, including computed history_link and agentic_workflow_url
   const templateContext = toSnakeCase({ ...ctx, historyLink, agenticWorkflowUrl, effectiveTokens, effectiveTokensFormatted, effectiveTokensSuffix });
@@ -386,7 +413,8 @@ function generateFooterWithMessages(workflowName, runUrl, workflowSource, workfl
   // Read effective tokens from environment variable if available.
   // GH_AW_EFFECTIVE_TOKENS is set by parse_mcp_gateway_log.cjs after computing ET
   // from the token-usage.jsonl produced by the firewall proxy.
-  const { effectiveTokens } = getEffectiveTokensFromEnv();
+  const modelName = process.env.GH_AW_ENGINE_MODEL || "";
+  const { effectiveTokens } = getEffectiveTokensFromEnv(modelName);
 
   // Read workflow emoji from environment variable if available.
   const emoji = process.env.GH_AW_WORKFLOW_EMOJI || undefined;