awslabs · herosjourney · Jul 22, 2026 · Jul 22, 2026 · Aug 13, 2026
@@ -685,7 +685,7 @@ The eval schema fields (validated, extras rejected):
   - `no_golden_cases: true` — when `total_cases == 0` (T2-2 abort / paste / vision-no-images / embeddings paths). Tells T2-6 to render "no quality data" instead of "0% pass rate".
   - Free-form addenda — manual-review notes, per-layer status hints. Keep total `notes` under ~500 chars to fit dashboards.
 - **`live_source_baseline`** — MANDATORY. `true` ONLY when §9 produced live side-by-side responses for at least one prompt. `false` when §9 was skipped (no source key) or every live call failed. The report banner depends on this flag.
-- **`judge_model`** — MANDATORY. Identifier of the LLM running THIS agent (e.g. `claude-opus-4-7`, `claude-haiku-4-5-20251001-v1:0`). The report discloses it so readers can assess same-family bias risk against the target Bedrock model. If you cannot identify the exact ID, pass `"unknown"` — never drop the field.
+- **`judge_model`** — MANDATORY. Identifier of the LLM running THIS agent (e.g. `claude-opus-4-8`, `claude-haiku-4-5-20251001-v1:0`). The report discloses it so readers can assess same-family bias risk against the target Bedrock model. If you cannot identify the exact ID, pass `"unknown"` — never drop the field.
 - **`source_baseline_quality`** — `'good'` (live baseline ran and looked fine), `'poor'` (live baseline ran but the source model's own output looked degraded — empty responses, error bodies, or obvious wrong-version behavior; per §9 step 3), or `'unknown'` (no live baseline ran). The orchestrator surfaces `'poor'` at the quality gate.
 
 `live_source_baseline` and `judge_model` are MANDATORY — always include both, every time, even if uncertain. Never omit them.
@@ -704,7 +704,7 @@ When §7 routed past §10–§13 with `total_cases == 0` (no golden dataset to e
   "failures": 1,
   "notes": "live_source_baseline_used_model: \n1 prompt needs manual review — see notes for details.",
   "live_source_baseline": true,
-  "judge_model": "claude-opus-4-7",
+  "judge_model": "claude-opus-4-8",
   "source_baseline_quality": "good"
 }
 ```
@@ -723,7 +723,7 @@ When `total_cases == 0` (T2-2 abort / paste / vision-no-images / embeddings path
   "failures": 0,
   "notes": "no_golden_cases: true\nreason: T2-2 reported total_golden_cases=0; layers 1/2 passed but no quality data to score.",
   "live_source_baseline": false,
-  "judge_model": "claude-opus-4-7",
+  "judge_model": "claude-opus-4-8",
   "source_baseline_quality": "unknown"
 }
 ```

@@ -334,7 +334,7 @@ <h3>High-level service mapping</h3>
           <tr><td>Cloud SQL PostgreSQL</td><td>RDS PostgreSQL db.t4g.micro</td><td><span class="badge-conf badge-tail">Tailored to your setup</span> (single-AZ)</td></tr>
           <tr><td>Cloud Storage</td><td>S3</td><td><span class="badge-conf badge-std">Standard pairing</span></td></tr>
           <tr><td>BigQuery</td><td>Deferred — specialist engagement</td><td><span class="badge-conf badge-defer">Specialist required</span></td></tr>
-          <tr><td>Vertex AI / Gemini</td><td>Bedrock Claude Sonnet 4.6</td><td><span class="badge-conf badge-tail">Tailored to your setup</span> (quality priority)</td></tr>
+          <tr><td>Vertex AI / Gemini</td><td>Bedrock Claude Sonnet 5</td><td><span class="badge-conf badge-tail">Tailored to your setup</span> (quality priority)</td></tr>
           <tr><td>Vertex AI Imagen</td><td>Bedrock Nova Canvas</td><td><span class="badge-conf badge-tail">Tailored to your setup</span></td></tr>
         </tbody>
       </table>
@@ -749,7 +749,7 @@ <h3>Model mappings</h3>
         <caption>GCP AI models mapped to Amazon Bedrock targets.</caption>
         <thead><tr><th scope="col">GCP model</th><th scope="col">Bedrock target</th><th scope="col">Workload</th><th scope="col">Complexity</th></tr></thead>
         <tbody>
-          <tr><td>gemini-2.5-flash</td><td>anthropic.claude-sonnet-4-6</td><td>Text generation, structured output</td><td>Medium</td></tr>
+          <tr><td>gemini-2.5-flash</td><td>anthropic.claude-sonnet-5</td><td>Text generation, structured output</td><td>Medium</td></tr>
           <tr><td>imagen-3.0-generate-001</td><td>amazon.nova-canvas-v1:0</td><td>Image generation (~500/mo)</td><td>Medium</td></tr>
         </tbody>
       </table>
@@ -759,13 +759,13 @@ <h3>Model cost comparison</h3>
         <caption>Bedrock model options vs current GCP AI spend ($1,250/mo).</caption>
         <thead><tr><th scope="col">Bedrock model</th><th scope="col">Monthly est.</th><th scope="col">vs GCP ($1,250/mo)</th><th scope="col">Quality</th><th scope="col">Notes</th></tr></thead>
         <tbody>
-          <tr><td><strong>Claude Sonnet 4.6 (recommended)</strong></td><td>$780</td><td class="savings">-38%</td><td>High</td><td>Quality priority (Q16); selected target</td></tr>
+          <tr><td><strong>Claude Sonnet 5 (recommended)</strong></td><td>$780</td><td class="savings">-38%</td><td>High</td><td>Quality priority (Q16); selected target</td></tr>
           <tr><td>Claude Haiku 4.5</td><td>$260</td><td class="savings">-79%</td><td>Medium-High</td><td>Balance of cost and quality</td></tr>
           <tr><td>Nova Lite v1</td><td>$14</td><td class="savings">-99%</td><td>Medium</td><td>Backup if A/B test passes quality bar</td></tr>
         </tbody>
       </table>
 
-      <h3>Recommended model token breakdown (Claude Sonnet 4.6)</h3>
+      <h3>Recommended model token breakdown (Claude Sonnet 5)</h3>
       <table>
         <caption>Monthly token volume and cost for the recommended model.</caption>
         <thead><tr><th scope="col">Component</th><th scope="col">Volume</th><th scope="col">Cost</th></tr></thead>
@@ -791,7 +791,7 @@ <h3>Code changes</h3>
 
       <h3>Production readiness checklist</h3>
       <ul class="compact">
-        <li>Bedrock model access enabled for Claude Sonnet 4.6 and Nova Canvas in us-west-2</li>
+        <li>Bedrock model access enabled for Claude Sonnet 5 and Nova Canvas in us-west-2</li>
         <li>IAM role with <code>bedrock:InvokeModel</code> on ECS task role</li>
         <li>Provider adapter deployed with <code>AI_PROVIDER</code> feature flag</li>
         <li>A/B test with &ge;100 representative prompts; image quality human evaluation</li>
@@ -849,7 +849,7 @@ <h2>Your Configuration Choices</h2>
             <td>What matters most for your AI workloads?</td>
             <td>Quality</td>
             <td>User answer</td>
-            <td>Claude Sonnet 4.6 recommended over lower-cost models</td>
+            <td>Claude Sonnet 5 recommended over lower-cost models</td>
           </tr>
           <tr>
             <td>Detected: Cloud SQL ZONAL &rarr; single-AZ availability</td>

@@ -61,7 +61,7 @@ the same topology discipline as Path 1:
 - **Primary request flow (solid edges):** User → the app's entry surface (the runtime hosting
   the migrated app + its `/invocations` entrypoint) → the app's orchestration layer (e.g.
   LangChain) → the Bedrock target model. Show the provider swap explicitly — label the model
-  node with the migrated target (e.g. "Bedrock: Claude Sonnet 4.6") and note the source it
+  node with the migrated target (e.g. "Bedrock: Claude Sonnet 5") and note the source it
   replaced where useful (e.g. "was: OpenAI gpt-3.5-turbo").
 - **State / memory (solid edge to a store):** if the app has conversation memory, show it as
   its own node mapped to its AWS target (e.g. in-process `ConversationBufferWindowMemory` →

@@ -66,7 +66,7 @@ summary first, technical detail after (single layered doc — do not fork by aud
 `$RUN_DIR/recommendation.md`. Append the freshness footer.
 
 For `migrate`: also fill Section 9 (Bedrock model) with the **coarse family mapping**
-(e.g. "GPT-4o → Claude Sonnet 4.6 family") and a note that detailed pricing/TCO come from the
+(e.g. "GPT-4o → Claude Sonnet 5 family") and a note that detailed pricing/TCO come from the
 migration plugins — no dollar figures. Section 10 (cost magnitude) presents the per-unit
 target-state bands from estimate.json and notes that the migration TCO comparison and
 current-spend delta are produced by the migration plugins.

@@ -126,7 +126,7 @@ contract but never replaces it. Resolve each unit's model and `api_path` from IT
 never take one unit's decision from another or from a run-wide plan scalar.
 
 For each unit, `design.json.units[<id>].model_recommendation.model` carries the selected
-path-specific model ID (for example, `anthropic.claude-sonnet-4-6`).
+path-specific model ID (for example, `anthropic.claude-sonnet-5`).
 `model_recommendation.invocation_model_id` is the exact account-callable ID only after CRIS
 resolution. These identities must remain separate.
 
@@ -155,11 +155,11 @@ The block's `target_bedrock_model` and path should equal `model_recommendation`
 validated them. If they differ, stop and return to Model Recommend; do not choose one silently.
 
 **Strip environment annotations from the model id.** The running assistant's model name may
-carry a context-window annotation like `[1m]` (e.g. `us.anthropic.claude-sonnet-4-6[1m]`).
+carry a context-window annotation like `[1m]` (e.g. `us.anthropic.claude-sonnet-5[1m]`).
 That suffix is a harness label, NOT a valid Bedrock model/inference-profile id — a Bedrock
 call with it 404s. Before writing the id into any generated file, strip any trailing
 `[...]` bracket annotation. The id you emit must be exactly what Bedrock accepts (e.g.
-`us.anthropic.claude-sonnet-4-6`). If unsure of the exact id, use the `TODO: verify model id`
+`us.anthropic.claude-sonnet-5`). If unsure of the exact id, use the `TODO: verify model id`
 placeholder rather than a bracket-tagged string.
 
 **Single-unit path:** when `units[]` length is 1, the above logic applies to the one unit —