diff --git a/.github/workflows/daily-cache-strategy-analyzer.lock.yml b/.github/workflows/daily-cache-strategy-analyzer.lock.yml
index 3b271e56f8c..625f9496b41 100644
--- a/.github/workflows/daily-cache-strategy-analyzer.lock.yml
+++ b/.github/workflows/daily-cache-strategy-analyzer.lock.yml
@@ -1080,6 +1080,7 @@ jobs:
           GH_AW_MAX_TURNS: ${{ vars.GH_AW_DEFAULT_MAX_TURNS || '' }}
           GH_AW_MCP_CONFIG: ${{ runner.temp }}/gh-aw/mcp-config/config.toml
           GH_AW_MODEL_AGENT_CODEX: ${{ needs.activation.outputs.model_size }}
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_AGENT_CODEX || vars.GH_AW_DEFAULT_MODEL_CODEX || 'gpt-5.4' }}
           GH_AW_PHASE: agent
           GH_AW_PROMPT: /tmp/gh-aw/aw-prompts/prompt.txt
           GH_AW_SAFE_OUTPUTS: ${{ steps.set-runtime-paths.outputs.GH_AW_SAFE_OUTPUTS }}
@@ -1875,7 +1876,8 @@ jobs:
         uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
         env:
           GH_AW_EVALS_QUESTIONS: '[{"id":"cache_logs_analyzed","question":"Did the agent analyze agentic workflow logs for cache misses and misconfigured caches?"},{"id":"issue_created_or_noop","question":"Was an issue created for cache problems or improvements, or was noop used when no problems were found?"}]'
-          GH_AW_EVALS_MODEL: "${{ needs.activation.outputs.model_size }}"
+          GH_AW_EVALS_MODEL: ${{ needs.activation.outputs.model_size }}
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_DETECTION_CODEX || vars.GH_AW_DEFAULT_MODEL_CODEX || 'gpt-5.4' }}
           GH_AW_EVALS_PHASE: setup
         with:
           script: |
@@ -2010,6 +2012,7 @@ jobs:
           GH_AW_MAX_TURNS: ${{ vars.GH_AW_DEFAULT_MAX_TURNS || '' }}
           GH_AW_MCP_CONFIG: ${{ runner.temp }}/gh-aw/mcp-config/config.toml
           GH_AW_MODEL_DETECTION_CODEX: ${{ needs.activation.outputs.model_size }}
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_DETECTION_CODEX || vars.GH_AW_DEFAULT_MODEL_CODEX || 'gpt-5.4' }}
           GH_AW_PHASE: detection
           GH_AW_PROMPT: /tmp/gh-aw/aw-prompts/prompt.txt
           GH_AW_VERSION: dev
@@ -2029,7 +2032,8 @@ jobs:
         uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
         env:
           GH_AW_EVALS_QUESTIONS: '[{"id":"cache_logs_analyzed","question":"Did the agent analyze agentic workflow logs for cache misses and misconfigured caches?"},{"id":"issue_created_or_noop","question":"Was an issue created for cache problems or improvements, or was noop used when no problems were found?"}]'
-          GH_AW_EVALS_MODEL: "${{ needs.activation.outputs.model_size }}"
+          GH_AW_EVALS_MODEL: ${{ needs.activation.outputs.model_size }}
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_DETECTION_CODEX || vars.GH_AW_DEFAULT_MODEL_CODEX || 'gpt-5.4' }}
           GH_AW_EVALS_PHASE: parse
           GITHUB_RUN_ID: ${{ github.run_id }}
         with:
diff --git a/.github/workflows/daily-caveman-optimizer.lock.yml b/.github/workflows/daily-caveman-optimizer.lock.yml
index 676b60720bf..dbe4ac6a902 100644
--- a/.github/workflows/daily-caveman-optimizer.lock.yml
+++ b/.github/workflows/daily-caveman-optimizer.lock.yml
@@ -983,6 +983,7 @@ jobs:
           GH_AW_LLM_PROVIDER: anthropic
           GH_AW_MAX_TURNS: ${{ vars.GH_AW_DEFAULT_MAX_TURNS || '' }}
           GH_AW_MCP_CONFIG: ${{ runner.temp }}/gh-aw/mcp-config/mcp-servers.json
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_AGENT_CLAUDE || vars.GH_AW_DEFAULT_MODEL_CLAUDE || '' }}
           GH_AW_PHASE: agent
           GH_AW_PROMPT: /tmp/gh-aw/aw-prompts/prompt.txt
           GH_AW_SAFE_OUTPUTS: ${{ steps.set-runtime-paths.outputs.GH_AW_SAFE_OUTPUTS }}
@@ -1762,7 +1763,8 @@ jobs:
         uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
         env:
           GH_AW_EVALS_QUESTIONS: '[{"id":"instruction_file_optimized","question":"Did the agent apply caveman optimization to an instruction file in .github/aw or .github/agents?"},{"id":"pr_created_or_noop","question":"Was a PR created with concise improvements, or was noop used when no improvements were needed?"}]'
-          GH_AW_EVALS_MODEL: "${{ needs.activation.outputs.model_size }}"
+          GH_AW_EVALS_MODEL: ${{ needs.activation.outputs.model_size }}
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_DETECTION_CLAUDE || vars.GH_AW_DEFAULT_MODEL_CLAUDE || '' }}
           GH_AW_EVALS_PHASE: setup
         with:
           script: |
@@ -1843,6 +1845,7 @@ jobs:
           DISABLE_TELEMETRY: 1
           GH_AW_LLM_PROVIDER: anthropic
           GH_AW_MAX_TURNS: ${{ vars.GH_AW_DEFAULT_MAX_TURNS || '' }}
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_DETECTION_CLAUDE || vars.GH_AW_DEFAULT_MODEL_CLAUDE || '' }}
           GH_AW_PHASE: detection
           GH_AW_PROMPT: /tmp/gh-aw/aw-prompts/prompt.txt
           GH_AW_VERSION: dev
@@ -1863,7 +1866,8 @@ jobs:
         uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
         env:
           GH_AW_EVALS_QUESTIONS: '[{"id":"instruction_file_optimized","question":"Did the agent apply caveman optimization to an instruction file in .github/aw or .github/agents?"},{"id":"pr_created_or_noop","question":"Was a PR created with concise improvements, or was noop used when no improvements were needed?"}]'
-          GH_AW_EVALS_MODEL: "${{ needs.activation.outputs.model_size }}"
+          GH_AW_EVALS_MODEL: ${{ needs.activation.outputs.model_size }}
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_DETECTION_CLAUDE || vars.GH_AW_DEFAULT_MODEL_CLAUDE || '' }}
           GH_AW_EVALS_PHASE: parse
           GITHUB_RUN_ID: ${{ github.run_id }}
         with:
diff --git a/.github/workflows/daily-doc-healer.lock.yml b/.github/workflows/daily-doc-healer.lock.yml
index b792f7b71ad..4543496fcc4 100644
--- a/.github/workflows/daily-doc-healer.lock.yml
+++ b/.github/workflows/daily-doc-healer.lock.yml
@@ -1087,6 +1087,7 @@ jobs:
           GH_AW_LLM_PROVIDER: anthropic
           GH_AW_MAX_TURNS: ${{ vars.GH_AW_DEFAULT_MAX_TURNS || '' }}
           GH_AW_MCP_CONFIG: ${{ runner.temp }}/gh-aw/mcp-config/mcp-servers.json
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_AGENT_CLAUDE || vars.GH_AW_DEFAULT_MODEL_CLAUDE || '' }}
           GH_AW_PHASE: agent
           GH_AW_PROMPT: /tmp/gh-aw/aw-prompts/prompt.txt
           GH_AW_SAFE_OUTPUTS: ${{ steps.set-runtime-paths.outputs.GH_AW_SAFE_OUTPUTS }}
@@ -1869,7 +1870,8 @@ jobs:
         uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
         env:
           GH_AW_EVALS_QUESTIONS: '[{"id":"gaps-confirmed","question":"Did the workflow identify at least one confirmed documentation gap to fix, or correctly conclude that no actionable gap remained?"},{"id":"pr-issue-or-noop","question":"Was a documentation pull request or issue created for confirmed gaps, or was noop used appropriately when nothing required action?"}]'
-          GH_AW_EVALS_MODEL: "${{ needs.activation.outputs.model_size }}"
+          GH_AW_EVALS_MODEL: ${{ needs.activation.outputs.model_size }}
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_DETECTION_CLAUDE || vars.GH_AW_DEFAULT_MODEL_CLAUDE || '' }}
           GH_AW_EVALS_PHASE: setup
         with:
           script: |
@@ -1950,6 +1952,7 @@ jobs:
           DISABLE_TELEMETRY: 1
           GH_AW_LLM_PROVIDER: anthropic
           GH_AW_MAX_TURNS: ${{ vars.GH_AW_DEFAULT_MAX_TURNS || '' }}
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_DETECTION_CLAUDE || vars.GH_AW_DEFAULT_MODEL_CLAUDE || '' }}
           GH_AW_PHASE: detection
           GH_AW_PROMPT: /tmp/gh-aw/aw-prompts/prompt.txt
           GH_AW_VERSION: dev
@@ -1970,7 +1973,8 @@ jobs:
         uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
         env:
           GH_AW_EVALS_QUESTIONS: '[{"id":"gaps-confirmed","question":"Did the workflow identify at least one confirmed documentation gap to fix, or correctly conclude that no actionable gap remained?"},{"id":"pr-issue-or-noop","question":"Was a documentation pull request or issue created for confirmed gaps, or was noop used appropriately when nothing required action?"}]'
-          GH_AW_EVALS_MODEL: "${{ needs.activation.outputs.model_size }}"
+          GH_AW_EVALS_MODEL: ${{ needs.activation.outputs.model_size }}
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_DETECTION_CLAUDE || vars.GH_AW_DEFAULT_MODEL_CLAUDE || '' }}
           GH_AW_EVALS_PHASE: parse
           GITHUB_RUN_ID: ${{ github.run_id }}
         with:
diff --git a/.github/workflows/test-quality-sentinel.lock.yml b/.github/workflows/test-quality-sentinel.lock.yml
index fa095fd4668..40a34231807 100644
--- a/.github/workflows/test-quality-sentinel.lock.yml
+++ b/.github/workflows/test-quality-sentinel.lock.yml
@@ -985,6 +985,7 @@ jobs:
           GH_AW_LLM_PROVIDER: github
           GH_AW_MAX_AI_CREDITS: ${{ vars.GH_AW_DEFAULT_MAX_AI_CREDITS || '1000' }}
           GH_AW_MAX_TURNS: ${{ vars.GH_AW_DEFAULT_MAX_TURNS || '' }}
+          GH_AW_MODEL_FALLBACK: ${{ vars.GH_AW_MODEL_AGENT_COPILOT || vars.GH_AW_DEFAULT_MODEL_COPILOT || 'claude-sonnet-4.6' }}
           GH_AW_PHASE: agent
           GH_AW_PROMPT: /tmp/gh-aw/aw-prompts/prompt.txt
           GH_AW_SAFE_OUTPUTS: ${{ steps.set-runtime-paths.outputs.GH_AW_SAFE_OUTPUTS }}
diff --git a/actions/setup/js/claude_harness.cjs b/actions/setup/js/claude_harness.cjs
index e84e02b496a..33be2194615 100644
--- a/actions/setup/js/claude_harness.cjs
+++ b/actions/setup/js/claude_harness.cjs
@@ -53,6 +53,7 @@ const { emitMissingToolPermissionIssue, hasExpectedSafeOutputs, hasNoopInSafeOut
 const { countPermissionDeniedIssues, hasNumerousPermissionDeniedIssues, extractDeniedCommands, buildMissingToolPermissionIssuePayload } = require("./permission_denied_helpers.cjs");
 const { detectNonRetryableHarnessGuard, buildSoftTimeoutGuard, emitSoftTimeoutSignal, isAuthenticationFailedError } = require("./harness_retry_guard.cjs");
 const { MODEL_NOT_SUPPORTED_PATTERN: INVALID_MODEL_ERROR_PATTERN } = require("./detect_agent_errors.cjs");
+const { applyModelFallback } = require("./model_fallback.cjs");
 
 // Pattern to detect Anthropic API overload errors (HTTP 529).
 // Matches "overloaded_error" from the Anthropic error type field, and the
@@ -318,6 +319,7 @@ function stripContinueArgs(args) {
  */
 async function buildClaudeChildEnv() {
   const childEnv = { ...process.env };
+  applyModelFallback(childEnv, "ANTHROPIC_MODEL", log);
   const provider = normalizeReflectProviderName(process.env.GH_AW_LLM_PROVIDER, "anthropic");
   try {
     const raw = fs.readFileSync(AWF_REFLECT_OUTPUT_PATH, "utf8");
@@ -611,6 +613,7 @@ if (typeof module !== "undefined" && module.exports) {
     hasExpectedSafeOutputs,
     resolveRetryConfig,
     resolveStartupRetryLimit,
+    applyModelFallback,
   };
 }
 
diff --git a/actions/setup/js/codex_harness.cjs b/actions/setup/js/codex_harness.cjs
index 92f97cc9d84..81fba7e97f9 100644
--- a/actions/setup/js/codex_harness.cjs
+++ b/actions/setup/js/codex_harness.cjs
@@ -54,6 +54,7 @@ const { countPermissionDeniedIssues, hasNumerousPermissionDeniedIssues, extractD
 const { detectNonRetryableHarnessGuard, buildSoftTimeoutGuard, emitSoftTimeoutSignal, isAuthenticationFailedError } = require("./harness_retry_guard.cjs");
 const { MODEL_NOT_SUPPORTED_PATTERN: INVALID_MODEL_ERROR_PATTERN } = require("./detect_agent_errors.cjs");
 const { resolveRetryConfig } = require("./harness_retry_config.cjs");
+const { applyModelFallback, injectModelFlagAfterExec } = require("./model_fallback.cjs");
 
 // Pattern to detect OpenAI rate-limit errors.
 // Matches the JSON error type field ("rate_limit_exceeded"), the HTTP status code
@@ -221,6 +222,16 @@ function injectJsonFlag(args) {
   return ["exec", "--json", ...args.slice(1)];
 }
 
+function getCodexModelEnvVar(env = process.env) {
+  if ("GH_AW_MODEL_DETECTION_CODEX" in env) {
+    return "GH_AW_MODEL_DETECTION_CODEX";
+  }
+  if ("GH_AW_MODEL_AGENT_CODEX" in env) {
+    return "GH_AW_MODEL_AGENT_CODEX";
+  }
+  return "";
+}
+
 /**
  * Build child process environment for Codex execution.
  * Preserve API keys captured at harness startup, even if the parent environment
@@ -425,6 +436,10 @@ async function main() {
     process.exit(1);
   }
 
+  const codexModelEnvVar = getCodexModelEnvVar(process.env);
+  const resolvedModel = codexModelEnvVar ? applyModelFallback(process.env, codexModelEnvVar, log) : "";
+  resolvedArgs = injectModelFlagAfterExec(resolvedArgs, resolvedModel);
+
   // Safe arg list for logging: when --prompt-file was present, the last element of
   // resolvedArgs is the resolved prompt content. Replace it with a placeholder so that
   // task instructions are never written to stderr or captured in agent logs.
@@ -641,6 +656,9 @@ if (typeof module !== "undefined" && module.exports) {
     hasNoopInSafeOutputs,
     hasExpectedSafeOutputs,
     resolveRetryConfig,
+    applyModelFallback,
+    injectModelFlagAfterExec,
+    getCodexModelEnvVar,
   };
 }
 
diff --git a/actions/setup/js/copilot_harness.cjs b/actions/setup/js/copilot_harness.cjs
index 4ed31e07d16..d29e70d4d44 100644
--- a/actions/setup/js/copilot_harness.cjs
+++ b/actions/setup/js/copilot_harness.cjs
@@ -65,6 +65,7 @@ const { runSafeOutputsCLI, buildMissingToolAlternatives, emitMissingToolPermissi
 const { countPermissionDeniedIssues, hasNumerousPermissionDeniedIssues, extractDeniedCommands, buildMissingToolPermissionIssuePayload } = require("./permission_denied_helpers.cjs");
 const { detectNonRetryableHarnessGuard, buildSoftTimeoutGuard, emitSoftTimeoutSignal, isAuthenticationFailedError: isCommonAuthenticationFailedError } = require("./harness_retry_guard.cjs");
 const { isCAPIQuotaExceededError } = require("./detect_agent_errors.cjs");
+const { applyModelFallback } = require("./model_fallback.cjs");
 const { loadModelsJson } = require("./model_costs.cjs");
 const { resolveConfiguredCopilotModel } = require("./resolve_model_alias.cjs");
 
@@ -950,6 +951,7 @@ async function main() {
     }
   }
 
+  applyModelFallback(process.env, "COPILOT_MODEL", log);
   applyCopilotModelAliasResolution({ awfReflectData, logger: log });
   applyCopilotWireAPI({ modelsJson: loadModelsJson(), logger: log });
 
@@ -1456,6 +1458,7 @@ if (typeof module !== "undefined" && module.exports) {
     parseCopilotSDKServerArgsFromEnv,
     isCAPIQuotaExceededError,
     hasTerminalSafeOutput,
+    applyModelFallback,
     applyCopilotModelAliasResolution,
     applyCopilotWireAPI,
     loadAwfConfigData,
diff --git a/actions/setup/js/copilot_sdk_driver.cjs b/actions/setup/js/copilot_sdk_driver.cjs
index 2e609e0462a..9716aaf3ffb 100644
--- a/actions/setup/js/copilot_sdk_driver.cjs
+++ b/actions/setup/js/copilot_sdk_driver.cjs
@@ -30,6 +30,7 @@ const fs = require("fs");
 const { runWithCopilotSDK, extractPromptFromArgs } = require("./copilot_sdk_session.cjs");
 const { parsePermissionConfigFromServerArgs } = require("./copilot_sdk_permissions.cjs");
 const { parseMultiProviderJson } = require("./copilot_sdk_multi_provider.cjs");
+const { applyModelFallback } = require("./model_fallback.cjs");
 
 // Re-export the session and permission helpers so that existing callers that
 // require("./copilot_sdk_driver.cjs") (e.g. copilot_harness.cjs) continue to work.
@@ -103,7 +104,7 @@ async function main() {
   const providers = multiProviderConfig.providers;
   /** @type {import("@github/copilot-sdk").ProviderModelConfig[]} */
   const sdkModels = multiProviderConfig.models;
-  let model = process.env.COPILOT_MODEL || multiProviderConfig.model || undefined;
+  let model = applyModelFallback(process.env, "COPILOT_MODEL", log) || multiProviderConfig.model || undefined;
   log(`multi-provider mode: ${providers.length} providers, ${sdkModels.length} models, model=${model ?? "(env)"}`);
   for (const p of providers) {
     log(`  provider: name=${p.name} type=${p.type} baseUrl=${p.baseUrl}${p.wireApi ? ` wireApi=${p.wireApi}` : ""}`);
diff --git a/actions/setup/js/copilot_sdk_session.cjs b/actions/setup/js/copilot_sdk_session.cjs
index 30c3a6251e9..1c0172ddaf1 100644
--- a/actions/setup/js/copilot_sdk_session.cjs
+++ b/actions/setup/js/copilot_sdk_session.cjs
@@ -36,6 +36,7 @@ const fs = require("fs");
 const path = require("path");
 const os = require("os");
 const { buildCopilotSDKPermissionHandler, getEnvPositiveIntOrDefault, parseMaxToolDenialsLimit, MAX_TOOL_DENIALS_DEFAULT } = require("./copilot_sdk_permissions.cjs");
+const { resolveModelWithFallback } = require("./model_fallback.cjs");
 const { extractShellCommandFromToolData } = require("./tool_call_details.cjs");
 
 // Default timeout for a single sendAndWait call: 10 minutes.
@@ -254,7 +255,7 @@ async function runWithCopilotSDK({ sdkUri, prompt, logger, attempt = 0, model, c
     // Build session config using the multi-provider surface.
     /** @type {import("@github/copilot-sdk").SessionConfig} */
     const sessionConfig = {
-      model: model || process.env.COPILOT_MODEL || undefined,
+      model: model || resolveModelWithFallback(process.env, "COPILOT_MODEL") || undefined,
       providers,
       models: providerModels,
       onPermissionRequest,
diff --git a/actions/setup/js/model_fallback.cjs b/actions/setup/js/model_fallback.cjs
new file mode 100644
index 00000000000..70c1f205ff2
--- /dev/null
+++ b/actions/setup/js/model_fallback.cjs
@@ -0,0 +1,48 @@
+"use strict";
+
+const MODEL_FALLBACK_ENV_VAR = "GH_AW_MODEL_FALLBACK";
+
+function readTrimmedEnv(env, name) {
+  return typeof env?.[name] === "string" ? env[name].trim() : "";
+}
+
+function resolveModelWithFallback(env, primaryEnvVar) {
+  return readTrimmedEnv(env, primaryEnvVar) || readTrimmedEnv(env, MODEL_FALLBACK_ENV_VAR);
+}
+
+/**
+ * @param {NodeJS.ProcessEnv} env
+ * @param {string} primaryEnvVar
+ * @param {(message: string) => void} [logger]
+ * @returns {string}
+ */
+function applyModelFallback(env, primaryEnvVar, logger = () => {}) {
+  const primary = readTrimmedEnv(env, primaryEnvVar);
+  if (primary) {
+    return primary;
+  }
+  const fallback = readTrimmedEnv(env, MODEL_FALLBACK_ENV_VAR);
+  if (fallback) {
+    env[primaryEnvVar] = fallback;
+    logger(`applied ${MODEL_FALLBACK_ENV_VAR} to ${primaryEnvVar}`);
+  }
+  return fallback;
+}
+
+function injectModelFlagAfterExec(args, model) {
+  if (!model || args.includes("--model")) {
+    return args;
+  }
+  const execIndex = args.indexOf("exec");
+  if (execIndex === -1) {
+    return [...args, "--model", model];
+  }
+  return [...args.slice(0, execIndex + 1), "--model", model, ...args.slice(execIndex + 1)];
+}
+
+module.exports = {
+  MODEL_FALLBACK_ENV_VAR,
+  resolveModelWithFallback,
+  applyModelFallback,
+  injectModelFlagAfterExec,
+};
diff --git a/actions/setup/js/model_fallback.test.js b/actions/setup/js/model_fallback.test.js
new file mode 100644
index 00000000000..ce524c81572
--- /dev/null
+++ b/actions/setup/js/model_fallback.test.js
@@ -0,0 +1,25 @@
+import { describe, expect, it } from "vitest";
+import { createRequire } from "module";
+
+const require = createRequire(import.meta.url);
+const { MODEL_FALLBACK_ENV_VAR, resolveModelWithFallback, applyModelFallback, injectModelFlagAfterExec } = require("./model_fallback.cjs");
+
+describe("model_fallback.cjs", () => {
+  it("prefers the primary model env var when it is non-empty", () => {
+    expect(resolveModelWithFallback({ COPILOT_MODEL: "gpt-5", [MODEL_FALLBACK_ENV_VAR]: "claude-sonnet-4.6" }, "COPILOT_MODEL")).toBe("gpt-5");
+  });
+
+  it("uses the fallback model when the primary resolves empty", () => {
+    const env = { COPILOT_MODEL: "", [MODEL_FALLBACK_ENV_VAR]: "claude-sonnet-4.6" };
+    expect(applyModelFallback(env, "COPILOT_MODEL")).toBe("claude-sonnet-4.6");
+    expect(env.COPILOT_MODEL).toBe("claude-sonnet-4.6");
+  });
+
+  it("injects --model immediately after exec when needed", () => {
+    expect(injectModelFlagAfterExec(["exec", "--json", "prompt"], "gpt-5")).toEqual(["exec", "--model", "gpt-5", "--json", "prompt"]);
+  });
+
+  it("does not inject --model when already present", () => {
+    expect(injectModelFlagAfterExec(["exec", "--model", "gpt-5", "--json"], "claude-sonnet-4.6")).toEqual(["exec", "--model", "gpt-5", "--json"]);
+  });
+});
diff --git a/actions/setup/js/run_evals.cjs b/actions/setup/js/run_evals.cjs
index f41af3a012e..75f2627a205 100644
--- a/actions/setup/js/run_evals.cjs
+++ b/actions/setup/js/run_evals.cjs
@@ -35,6 +35,7 @@ const path = require("path");
 
 const { ERR_VALIDATION } = require("./error_codes.cjs");
 const { EVALS_OUTPUT_PATH } = require("./evals_constants.cjs");
+const { resolveModelWithFallback } = require("./model_fallback.cjs");
 
 const EVALS_DIR = "/tmp/gh-aw/evals";
 const EVALS_LOG_PATH = "/tmp/gh-aw/evals/evals.log";
@@ -118,7 +119,7 @@ async function setupMain() {
  */
 async function parseMain() {
   const questionsRaw = process.env.GH_AW_EVALS_QUESTIONS;
-  const model = process.env.GH_AW_EVALS_MODEL || "";
+  const model = resolveModelWithFallback(process.env, "GH_AW_EVALS_MODEL") || "";
   const runID = process.env.GITHUB_RUN_ID || "unknown";
 
   /** @type {Array<{id: string, question: string}>} */
diff --git a/actions/setup/js/run_evals.test.cjs b/actions/setup/js/run_evals.test.cjs
index 7bbf2c4ea9b..52825d72464 100644
--- a/actions/setup/js/run_evals.test.cjs
+++ b/actions/setup/js/run_evals.test.cjs
@@ -6,6 +6,7 @@ const EVALS_DIR = "/tmp/gh-aw/evals";
 const EVALS_LOG_PATH = `${EVALS_DIR}/evals.log`;
 const EVALS_OUTPUT_PATH = "/tmp/gh-aw/evals.jsonl";
 const require = createRequire(import.meta.url);
+const { MODEL_FALLBACK_ENV_VAR } = require("./model_fallback.cjs");
 const { parseMain, extractAssistantTextFromJsonlLog } = require("./run_evals.cjs");
 
 const mockCore = {
@@ -75,6 +76,19 @@ describe("run_evals.cjs", () => {
     expect(JSON.parse(line).runid).toBe("unknown");
   });
 
+  it("uses GH_AW_MODEL_FALLBACK when GH_AW_EVALS_MODEL resolves empty", async () => {
+    vi.stubEnv("GH_AW_EVALS_QUESTIONS", JSON.stringify([{ id: "labels-applied", question: "Did labels get applied?" }]));
+    vi.stubEnv("GH_AW_EVALS_MODEL", "");
+    vi.stubEnv(MODEL_FALLBACK_ENV_VAR, "claude-sonnet-4.6");
+    vi.stubEnv("GITHUB_RUN_ID", "123456789");
+    fs.writeFileSync(EVALS_LOG_PATH, "labels-applied: YES\n", "utf8");
+
+    await parseMain();
+
+    const [line] = fs.readFileSync(EVALS_OUTPUT_PATH, "utf8").trim().split("\n");
+    expect(JSON.parse(line).model).toBe("claude-sonnet-4.6");
+  });
+
   it("parses answers from Pi v3 JSONL turn_end events (positional format)", async () => {
     vi.stubEnv(
       "GH_AW_EVALS_QUESTIONS",
diff --git a/pkg/constants/engine_constants.go b/pkg/constants/engine_constants.go
index fdc66a30b7c..ef63b38a4d2 100644
--- a/pkg/constants/engine_constants.go
+++ b/pkg/constants/engine_constants.go
@@ -243,6 +243,11 @@ const (
 	// EnvVarModelAgentPi configures the default Pi model for agent execution
 	EnvVarModelAgentPi = "GH_AW_MODEL_AGENT_PI"
 
+	// EnvVarModelFallback carries the standard org/enterprise/built-in model fallback
+	// expression for the current engine step. Runtime JavaScript harnesses can use this
+	// when a configured model expression resolves to an empty string.
+	EnvVarModelFallback = "GH_AW_MODEL_FALLBACK"
+
 	// CopilotCLIModelEnvVar is the native environment variable name supported by the Copilot CLI
 	// for selecting the model. Setting this env var is equivalent to passing --model to the CLI.
 	CopilotCLIModelEnvVar = "COPILOT_MODEL"
diff --git a/pkg/workflow/claude_engine.go b/pkg/workflow/claude_engine.go
index 8cf2070a4e7..0a361135e0a 100644
--- a/pkg/workflow/claude_engine.go
+++ b/pkg/workflow/claude_engine.go
@@ -474,6 +474,16 @@ func (e *ClaudeEngine) GetExecutionSteps(workflowData *WorkflowData, logFile str
 	// When model is not configured, fall back to GH_AW_MODEL_AGENT/DETECTION_CLAUDE so users
 	// can set a default via GitHub Actions variables.
 	if modelConfigured {
+		isDetectionJob := workflowData.SafeOutputs == nil
+		var claudeModelVar string
+		if isDetectionJob {
+			claudeModelVar = constants.EnvVarModelDetectionClaude
+		} else {
+			claudeModelVar = constants.EnvVarModelAgentClaude
+		}
+		if containsExpression(workflowData.Model) {
+			env[constants.EnvVarModelFallback] = compilerenv.BuildModelOverrideExpressionEmptyFallback(claudeModelVar, compilerenv.DefaultModelClaude)
+		}
 		claudeLog.Printf("Setting %s env var for model: %s", constants.ClaudeCLIModelEnvVar, workflowData.Model)
 		env[constants.ClaudeCLIModelEnvVar] = workflowData.Model
 	} else {
diff --git a/pkg/workflow/codex_engine.go b/pkg/workflow/codex_engine.go
index 9f53295235a..8876100600b 100644
--- a/pkg/workflow/codex_engine.go
+++ b/pkg/workflow/codex_engine.go
@@ -487,6 +487,9 @@ mkdir -p "$CODEX_HOME/logs"
 	// When model is configured (static or GitHub Actions expression), set the env var directly.
 	// When not configured, use the GitHub variable fallback so users can set a default.
 	if modelConfigured {
+		if containsExpression(workflowData.Model) {
+			env[constants.EnvVarModelFallback] = compilerenv.BuildModelOverrideExpression(modelEnvVar, compilerenv.DefaultModelCodex, constants.CodexDefaultModel)
+		}
 		codexEngineLog.Printf("Setting %s env var for model: %s", modelEnvVar, workflowData.Model)
 		env[modelEnvVar] = workflowData.Model
 	} else {
diff --git a/pkg/workflow/copilot_engine_execution.go b/pkg/workflow/copilot_engine_execution.go
index 83dac99da39..005fcfede45 100644
--- a/pkg/workflow/copilot_engine_execution.go
+++ b/pkg/workflow/copilot_engine_execution.go
@@ -588,6 +588,9 @@ func (e *CopilotEngine) addCopilotModelEnv(env map[string]string, workflowData *
 	// The model is always passed via the native COPILOT_MODEL env var, which the Copilot CLI reads directly.
 	// When model is not configured, map the GitHub org variable to COPILOT_MODEL so users can set a default.
 	if modelConfigured {
+		if containsExpression(workflowData.Model) {
+			env[constants.EnvVarModelFallback] = compilerenv.BuildModelOverrideExpression(modelEnvVar, compilerenv.DefaultModelCopilot, constants.CopilotBYOKDefaultModel)
+		}
 		copilotExecLog.Printf("Setting %s env var for model: %s", constants.CopilotCLIModelEnvVar, workflowData.Model)
 		env[constants.CopilotCLIModelEnvVar] = workflowData.Model
 		return
diff --git a/pkg/workflow/evals_steps.go b/pkg/workflow/evals_steps.go
index 82cacd4af60..63a8f98c505 100644
--- a/pkg/workflow/evals_steps.go
+++ b/pkg/workflow/evals_steps.go
@@ -121,11 +121,13 @@ await main();`
 		fmt.Sprintf("        uses: %s\n", getCachedActionPin("actions/github-script", data)),
 		"        env:\n",
 		fmt.Sprintf("          GH_AW_EVALS_QUESTIONS: '%s'\n", escapeYAMLSingleQuoted(questionsJSON)),
-		fmt.Sprintf("          GH_AW_EVALS_MODEL: %q\n", model),
+	}
+	steps = appendEvalsModelEnvLines(steps, c.getEvalsEngineID(data), model)
+	steps = append(steps,
 		"          GH_AW_EVALS_PHASE: setup\n",
 		"        with:\n",
 		"          script: |\n",
-	}
+	)
 	steps = append(steps, FormatJavaScriptForYAML(script)...)
 	return steps
 }
@@ -277,12 +279,14 @@ await main();`
 		fmt.Sprintf("        uses: %s\n", getCachedActionPin("actions/github-script", data)),
 		"        env:\n",
 		fmt.Sprintf("          GH_AW_EVALS_QUESTIONS: '%s'\n", escapeYAMLSingleQuoted(questionsJSON)),
-		fmt.Sprintf("          GH_AW_EVALS_MODEL: %q\n", model),
+	}
+	steps = appendEvalsModelEnvLines(steps, c.getEvalsEngineID(data), model)
+	steps = append(steps,
 		"          GH_AW_EVALS_PHASE: parse\n",
 		"          GITHUB_RUN_ID: ${{ github.run_id }}\n",
 		"        with:\n",
 		"          script: |\n",
-	}
+	)
 	steps = append(steps, FormatJavaScriptForYAML(script)...)
 	return steps
 }
@@ -408,6 +412,35 @@ func (c *Compiler) resolveEvalsExecutionModel(data *WorkflowData) string {
 	return model
 }
 
+func appendEvalsModelEnvLines(steps []string, engineID, model string) []string {
+	hasExpression := containsExpression(model)
+	modelValue := fmt.Sprintf("%q", model)
+	if hasExpression {
+		modelValue = model
+	}
+	steps = append(steps, fmt.Sprintf("          GH_AW_EVALS_MODEL: %s\n", modelValue))
+	if !hasExpression {
+		return steps
+	}
+	if fallback := buildEvalsModelFallbackExpression(engineID); fallback != "" {
+		steps = append(steps, fmt.Sprintf("          %s: %s\n", constants.EnvVarModelFallback, fallback))
+	}
+	return steps
+}
+
+func buildEvalsModelFallbackExpression(engineID string) string {
+	switch engineID {
+	case string(constants.CopilotEngine):
+		return compilerenv.BuildModelOverrideExpression(constants.EnvVarModelDetectionCopilot, compilerenv.DefaultModelCopilot, constants.CopilotBYOKDefaultModel)
+	case string(constants.ClaudeEngine):
+		return compilerenv.BuildModelOverrideExpressionEmptyFallback(constants.EnvVarModelDetectionClaude, compilerenv.DefaultModelClaude)
+	case string(constants.CodexEngine):
+		return compilerenv.BuildModelOverrideExpression(constants.EnvVarModelDetectionCodex, compilerenv.DefaultModelCodex, constants.CodexDefaultModel)
+	default:
+		return ""
+	}
+}
+
 func (c *Compiler) collectEvalsSecretReferences(data *WorkflowData) []string {
 	if data.Evals == nil {
 		return nil
diff --git a/pkg/workflow/evals_steps_test.go b/pkg/workflow/evals_steps_test.go
index a834aaaab2b..63711ca310a 100644
--- a/pkg/workflow/evals_steps_test.go
+++ b/pkg/workflow/evals_steps_test.go
@@ -3,10 +3,12 @@
 package workflow
 
 import (
+	"fmt"
 	"strings"
 	"testing"
 
 	"github.com/github/gh-aw/pkg/constants"
+	"github.com/github/gh-aw/pkg/workflow/compilerenv"
 )
 
 // TestBuildEvalsEngineStepsArcDindTopology verifies that the evals job
@@ -198,6 +200,38 @@ func TestBuildParseEvalsResultsStepUsesResolvedExecutionModel(t *testing.T) {
 	}
 }
 
+func TestBuildParseEvalsResultsStepUsesExpressionModelAndFallbackEnv(t *testing.T) {
+	compiler := NewCompiler()
+
+	data := &WorkflowData{
+		AI:    "copilot",
+		Model: "${{ inputs.model }}",
+		EngineConfig: &EngineConfig{
+			ID: "copilot",
+		},
+		Evals: &EvalsConfig{
+			Questions: []EvalDefinition{
+				{ID: "builds", Question: "Does it build?"},
+			},
+		},
+	}
+
+	steps := strings.Join(compiler.buildParseEvalsResultsStep(data), "")
+	if !strings.Contains(steps, `GH_AW_EVALS_MODEL: ${{ inputs.model }}`) {
+		t.Errorf("expected parse step to preserve expression-backed evals model; got:\n%s", steps)
+	}
+	expectedFallbackEnvLine := fmt.Sprintf(
+		"%s: ${{ vars.%s || vars.%s || '%s' }}",
+		constants.EnvVarModelFallback,
+		constants.EnvVarModelDetectionCopilot,
+		compilerenv.DefaultModelCopilot,
+		constants.CopilotBYOKDefaultModel,
+	)
+	if !strings.Contains(steps, expectedFallbackEnvLine) {
+		t.Errorf("expected parse step to expose evals model fallback env %q; got:\n%s", expectedFallbackEnvLine, steps)
+	}
+}
+
 // TestBuildEvalsJobStepsCodexNoDuplicateContainerDownload verifies that a codex
 // workflow with evals does not generate a duplicate "Download container images" step.
 // Regression: buildEvalsJobSteps previously called buildPullAWFContainersStep
diff --git a/pkg/workflow/model_env_vars_test.go b/pkg/workflow/model_env_vars_test.go
index 9a048d657c2..21309bd260e 100644
--- a/pkg/workflow/model_env_vars_test.go
+++ b/pkg/workflow/model_env_vars_test.go
@@ -212,6 +212,9 @@ func TestExplicitModelConfigOverridesEnvVar(t *testing.T) {
 	if strings.Contains(stepsContent, constants.EnvVarModelAgentCopilot+":") {
 		t.Errorf("Fallback env var %s should not be present when model is explicitly configured", constants.EnvVarModelAgentCopilot)
 	}
+	if strings.Contains(stepsContent, constants.EnvVarModelFallback+":") {
+		t.Errorf("Fallback env var %s should not be present for a literal configured model", constants.EnvVarModelFallback)
+	}
 
 	// The model should be passed via the native COPILOT_MODEL env var (not via --model flag)
 	expectedEnvLine := constants.CopilotCLIModelEnvVar + ": gpt-4"
@@ -305,40 +308,54 @@ func TestExpressionModelUsesEnvVar(t *testing.T) {
 		name                 string
 		engine               string
 		model                string
-		expectedEnvVar       string
-		expectedEnvVal       string
+		expectedModelEnvVar  string
+		expectedModelEnvVal  string
+		expectedFallbackVal  string
 		expectShellExpansion bool // whether command should use ${VAR:+ --model "$VAR"}
 	}{
 		{
-			name:                 "Copilot agent with inputs.model expression uses native COPILOT_MODEL",
+			name:                 "Copilot agent keeps pure expression model and adds JS fallback env",
 			engine:               "copilot",
 			model:                "${{ inputs.model }}",
-			expectedEnvVar:       constants.CopilotCLIModelEnvVar,
-			expectedEnvVal:       "${{ inputs.model }}",
+			expectedModelEnvVar:  constants.CopilotCLIModelEnvVar,
+			expectedModelEnvVal:  "${{ inputs.model }}",
+			expectedFallbackVal:  "${{ vars." + constants.EnvVarModelAgentCopilot + " || vars." + compilerenv.DefaultModelCopilot + " || '" + constants.CopilotBYOKDefaultModel + "' }}",
 			expectShellExpansion: false, // Copilot reads COPILOT_MODEL natively, no shell expansion needed
 		},
 		{
-			name:                 "Copilot agent with vars.model expression uses native COPILOT_MODEL",
+			name:                 "Copilot agent keeps composite expression model and adds JS fallback env",
 			engine:               "copilot",
-			model:                "${{ vars.MY_MODEL }}",
-			expectedEnvVar:       constants.CopilotCLIModelEnvVar,
-			expectedEnvVal:       "${{ vars.MY_MODEL }}",
+			model:                "${{ inputs.provider }}/${{ inputs.model }}",
+			expectedModelEnvVar:  constants.CopilotCLIModelEnvVar,
+			expectedModelEnvVal:  "${{ inputs.provider }}/${{ inputs.model }}",
+			expectedFallbackVal:  "${{ vars." + constants.EnvVarModelAgentCopilot + " || vars." + compilerenv.DefaultModelCopilot + " || '" + constants.CopilotBYOKDefaultModel + "' }}",
 			expectShellExpansion: false,
 		},
 		{
-			name:                 "Claude agent with inputs.model expression uses native ANTHROPIC_MODEL",
+			name:                 "Claude agent keeps expression model and adds JS fallback env",
 			engine:               "claude",
 			model:                "${{ inputs.model }}",
-			expectedEnvVar:       constants.ClaudeCLIModelEnvVar,
-			expectedEnvVal:       "${{ inputs.model }}",
+			expectedModelEnvVar:  constants.ClaudeCLIModelEnvVar,
+			expectedModelEnvVal:  "${{ inputs.model }}",
+			expectedFallbackVal:  "${{ vars." + constants.EnvVarModelAgentClaude + " || vars." + compilerenv.DefaultModelClaude + " || '' }}",
 			expectShellExpansion: false, // Claude reads ANTHROPIC_MODEL natively, no shell expansion needed
 		},
 		{
-			name:                 "Codex agent with inputs.model expression",
+			name:                 "Claude agent keeps composite expression model and adds JS fallback env",
+			engine:               "claude",
+			model:                "${{ inputs.provider }}/${{ inputs.model }}",
+			expectedModelEnvVar:  constants.ClaudeCLIModelEnvVar,
+			expectedModelEnvVal:  "${{ inputs.provider }}/${{ inputs.model }}",
+			expectedFallbackVal:  "${{ vars." + constants.EnvVarModelAgentClaude + " || vars." + compilerenv.DefaultModelClaude + " || '' }}",
+			expectShellExpansion: false,
+		},
+		{
+			name:                 "Codex agent keeps composite expression model and adds JS fallback env",
 			engine:               "codex",
-			model:                "${{ inputs.model }}",
-			expectedEnvVar:       constants.EnvVarModelAgentCodex,
-			expectedEnvVal:       "${{ inputs.model }}",
+			model:                "${{ inputs.provider }}/${{ inputs.model }}",
+			expectedModelEnvVar:  constants.EnvVarModelAgentCodex,
+			expectedModelEnvVal:  "${{ inputs.provider }}/${{ inputs.model }}",
+			expectedFallbackVal:  "${{ vars." + constants.EnvVarModelAgentCodex + " || vars." + compilerenv.DefaultModelCodex + " || '" + constants.CodexDefaultModel + "' }}",
 			expectShellExpansion: true, // Codex has no native model env var, uses shell expansion
 		},
 	}
@@ -381,18 +398,22 @@ func TestExpressionModelUsesEnvVar(t *testing.T) {
 			}
 
 			// The env var must be set to the expression value
-			expectedEnvLine := tt.expectedEnvVar + ": " + tt.expectedEnvVal
-			if !strings.Contains(stepsContent, expectedEnvLine) {
-				t.Errorf("Expected env line '%s' not found in steps:\n%s", expectedEnvLine, stepsContent)
+			expectedModelEnvLine := tt.expectedModelEnvVar + ": " + tt.expectedModelEnvVal
+			if !strings.Contains(stepsContent, expectedModelEnvLine) {
+				t.Errorf("Expected env line '%s' not found in steps:\n%s", expectedModelEnvLine, stepsContent)
+			}
+			expectedFallbackEnvLine := constants.EnvVarModelFallback + ": " + tt.expectedFallbackVal
+			if !strings.Contains(stepsContent, expectedFallbackEnvLine) {
+				t.Errorf("Expected fallback env line '%s' not found in steps:\n%s", expectedFallbackEnvLine, stepsContent)
 			}
 
 			// Check shell expansion expectation
-			shellExpansionPattern := "${" + tt.expectedEnvVar + ":+"
+			shellExpansionPattern := "${" + tt.expectedModelEnvVar + ":+"
 			hasShellExpansion := strings.Contains(stepsContent, shellExpansionPattern)
 			if tt.expectShellExpansion && !hasShellExpansion {
-				t.Errorf("Expected conditional env var usage '${%s:+' not found in steps:\n%s", tt.expectedEnvVar, stepsContent)
+				t.Errorf("Expected conditional env var usage '${%s:+' not found in steps:\n%s", tt.expectedModelEnvVar, stepsContent)
 			} else if !tt.expectShellExpansion && hasShellExpansion {
-				t.Errorf("Unexpected conditional env var usage '${%s:+' found in steps (should use native env var):\n%s", tt.expectedEnvVar, stepsContent)
+				t.Errorf("Unexpected conditional env var usage '${%s:+' found in steps (should use native env var):\n%s", tt.expectedModelEnvVar, stepsContent)
 			}
 		})
 	}
@@ -430,10 +451,14 @@ func TestExpressionModelDetectionJobUsesEnvVar(t *testing.T) {
 	}
 	stepsContent := stepsStr.String()
 
-	// Detection job for Copilot should use COPILOT_MODEL (native CLI env var)
-	expectedEnvLine := constants.CopilotCLIModelEnvVar + ": ${{ inputs.model }}"
-	if !strings.Contains(stepsContent, expectedEnvLine) {
-		t.Errorf("Expected env line '%s' not found in steps:\n%s", expectedEnvLine, stepsContent)
+	// Detection job for Copilot should keep the configured expression and expose the runtime fallback.
+	expectedModelEnvLine := constants.CopilotCLIModelEnvVar + ": ${{ inputs.model }}"
+	if !strings.Contains(stepsContent, expectedModelEnvLine) {
+		t.Errorf("Expected env line '%s' not found in steps:\n%s", expectedModelEnvLine, stepsContent)
+	}
+	expectedFallbackEnvLine := constants.EnvVarModelFallback + ": ${{ vars." + constants.EnvVarModelDetectionCopilot + " || vars." + compilerenv.DefaultModelCopilot + " || '" + constants.CopilotBYOKDefaultModel + "' }}"
+	if !strings.Contains(stepsContent, expectedFallbackEnvLine) {
+		t.Errorf("Expected fallback env line '%s' not found in steps:\n%s", expectedFallbackEnvLine, stepsContent)
 	}
 
 	// Must not embed expression directly in shell command