fix(ai_review): show actual model in cost line (was hardcoded glm-5.2:cloud)

The webhook pod now routes through headroom's MiniMax-M2.7 endpoint, but the
cost line in the AI-usage collapsible still read 'headroom glm-5.2:cloud'.
Resolve a single display_model at the top of review_pr (OPENCODE_MODEL env
wins, else headroom/{OLLAMA_MODEL}) and pass it to:
  * the opencode subprocess (was already doing this on the same line, now
    sharing the value)
  * format_review_body so REVIEW_HEADER also reflects the actual run
  * _render_collapsible_usage so the parenthetical reads
    '({display_model} — free tier)' or '({display_model} — billed)'.

Test additions in tests/pilot/test_ai_review.py cover:
  * the parenthetical picks up the passed-in model verbatim
  * the full provider prefix survives (headroom/<id>) for the opencode path
  * nonzero cost flips the inner clause from 'free tier' to 'billed'
  * the existing nonzero-cost assertion flips to assert 'billed' instead
    of dropping the parenthetical entirely

pilot/ai_review.py

@@ -1067,7 +1067,11 @@ def _render_collapsible_usage(usage: dict | None, model: str, config: dict | Non
 
     Empty string when `usage` is None. The cost-equivalent line is always
     shown (it's the operator's budgeting signal). The `actual` line is shown
-    but the FREE-TIER note is collapsed into a single short clause.
+    and its parenthetical clause reflects the *actually-routed* model —
+    typically `OPENCODE_MODEL` if set, otherwise `headroom/{OLLAMA_MODEL}` —
+    not a stale literal. `model` here is the resolved display name (the caller
+    computes it once and passes it everywhere: this helper, `format_review_body`,
+    and the opencode subprocess). Cost == 0 → "free tier"; nonzero → "billed".
     """
     if not usage:
         return ""
@@ -1075,7 +1079,7 @@ def _render_collapsible_usage(usage: dict | None, model: str, config: dict | Non
     dur_s = f"{dur}s" if dur is not None else "?"
     actual = usage.get("cost") or 0.0
     actual_s = f"${actual:.4f}" if actual else "$0.00"
-    actual_note = " (headroom glm-5.2:cloud — free tier)" if not actual else ""
+    actual_note = f" ({model} — {'free tier' if not actual else 'billed'})"
     price_key, price_err = _resolve_price_target(config)
     from cost_model import PRICES
     eq = equivalent_cost(usage, price_key)
@@ -1839,6 +1843,13 @@ def review_pr(
     Both the CI `run()` entry point and the central webhook server call this.
     """
     try:
+        # Resolve the model display name once and pass it to every consumer
+        # (opencode subprocess, REVIEW_HEADER, cost-line parenthetical). Env
+        # override wins; otherwise we prefix the OLLAMA_MODEL bare id with
+        # the headroom provider so the line reads `headroom/<model>` instead
+        # of a stale hardcoded literal.
+        display_model = os.environ.get("OPENCODE_MODEL") or f"headroom/{model}"
+
         reviews = fetch_existing_reviews(api, repo, index, token)
         # Dedupe: already reviewed this exact commit -> nothing to do.
         if sha and sha in reviewed_shas(reviews):
@@ -1847,7 +1858,7 @@ def review_pr(
 
         raw_diff, _truncated, _orig = fetch_pr_diff(api, repo, index, token, max_chars)
         if not raw_diff.strip():
-            post_review(api, repo, index, token, format_review_body("No diff content to review.", model, sha))
+            post_review(api, repo, index, token, format_review_body("No diff content to review.", display_model, sha))
             return True
 
         config = fetch_repo_config(api, repo, token, ref=base_ref)
@@ -1886,10 +1897,9 @@ def review_pr(
             # the brief, and the pragent agent factory; returns stdout with a
             # summary + findings JSON. We parse + anchor + post here.
             import opencode_review  # local import keeps the ollama path dep-free
-            # opencode wants a provider-prefixed model ref (headroom/glm-5.2:cloud);
-            # `model` here is the bare id (OLLAMA_MODEL). OPENCODE_MODEL overrides
-            # with the full ref; otherwise we prefix the configured provider.
-            oc_model = os.environ.get("OPENCODE_MODEL") or f"headroom/{model}"
+            # Reuse the display_model resolved above for the subprocess — same
+            # provider-prefixed ref goes to the engine and into the review body.
+            oc_model = display_model
             # Multi-lens fan-out: when the repo declared `reviewers[]` (or the
             # operator pinned PRAGENT_REVIEWERS=1), spawn one opencode subprocess
             # per lens in parallel and synthesize. Falls through to the legacy
@@ -1932,10 +1942,10 @@ def review_pr(
                     report_usage = pr_has_label(api, repo, index, token, AI_USAGE_LABEL)
                     if report_usage and usage and usage.get('output'):
                         compute_attribution(findings, usage['output'])
-                usage_section = _render_collapsible_usage(usage, model, config=config) if report_usage else ""
+                usage_section = _render_collapsible_usage(usage, display_model, config=config) if report_usage else ""
                 post_review(api, repo, index, token, format_review_body(
                     salvaged or "AI review produced no parseable output.",
-                    model, sha, usage_section=usage_section))
+                    display_model, sha, usage_section=usage_section))
                 return True
         else:
             user_prompt = build_user_prompt(title, body + compression_note, diff, config, prior, additional_context)
@@ -1979,7 +1989,7 @@ def review_pr(
             report_usage = pr_has_label(api, repo, index, token, AI_USAGE_LABEL)
             if report_usage and usage and usage.get('output'):
                 compute_attribution(findings, usage['output'])
-        usage_section = _render_collapsible_usage(usage, model, config=config) if report_usage else ""
+        usage_section = _render_collapsible_usage(usage, display_model, config=config) if report_usage else ""
 
         # Anchor against the RAW diff, never the compressed one. Compression
         # drops context lines, so a finding on a line that survived in the file
@@ -1998,7 +2008,7 @@ def review_pr(
         if bullets:
             summary_parts.append("### Unanchored Notes\n\n" + bullets)
         summary_body = format_review_body(
-            "\n\n".join(summary_parts), model, sha,
+            "\n\n".join(summary_parts), display_model, sha,
             summary=review_summary,
             usage_section=usage_section,
             summary_changes=summary_changes,
@@ -2016,7 +2026,7 @@ def review_pr(
         return True
     except Exception as e:  # fail-open
         try:
-            post_review(api, repo, index, token, format_review_body(f"⚠️ AI review failed: {e}", model, sha))
+            post_review(api, repo, index, token, format_review_body(f"⚠️ AI review failed: {e}", display_model, sha))
         except Exception as e2:
             print(f"pragent: could not post failure note: {e2}", file=sys.stderr)
         print(f"pragent: review failed: {e}", file=sys.stderr)

tests/pilot/test_ai_review.py

@@ -771,6 +771,39 @@ def test_render_collapsible_usage_cost_nonzero_drops_free_tier_note():
              "cache_write": 0, "total": 10, "cost": 0.0123, "steps": 1, "duration_s": 1.0}
     sec = _render_collapsible_usage(usage, "m", config=None)
     assert "$0.0123" in sec
+    # Was hardcoded "free tier" previously; now says "billed" since cost > 0.
+    assert "billed" in sec
+    assert "free tier" not in sec
+
+
+def test_render_collapsible_usage_uses_passed_model_for_free_tier_clause():
+    # Regression: the cost parenthetical must reflect the actually-routed model,
+    # not a stale hardcoded `headroom glm-5.2:cloud` literal that predates the
+    # MiniMax / Anthropic switch.
+    usage = {"input": 10, "output": 0, "reasoning": 0, "cache_read": 0,
+             "cache_write": 0, "total": 10, "cost": 0.0, "steps": 1, "duration_s": 1.0}
+    sec = _render_collapsible_usage(usage, "MiniMax-M2.7", config=None)
+    # The parenthetical clause is "(<model> — free tier)" — a model name MUST
+    # sit immediately before "— free tier".
+    assert "(MiniMax-M2.7 — free tier)" in sec
+    # And the stale hardcoded model name must no longer appear anywhere.
+    assert "glm-5.2:cloud" not in sec
+
+
+def test_render_collapsible_usage_full_provider_prefix_in_display():
+    # When the caller has resolved a provider-prefixed model ref (the opencode
+    # subprocess path), the parenthetical should mirror that verbatim.
+    usage = {"input": 10, "output": 0, "reasoning": 0, "cache_read": 0,
+             "cache_write": 0, "total": 10, "cost": 0.0, "steps": 1, "duration_s": 1.0}
+    sec = _render_collapsible_usage(usage, "headroom/MiniMax-M2.7", config=None)
+    assert "(headroom/MiniMax-M2.7 — free tier)" in sec
+
+
+def test_render_collapsible_usage_nonzero_cost_says_billed():
+    usage = {"input": 10, "output": 0, "reasoning": 0, "cache_read": 0,
+             "cache_write": 0, "total": 10, "cost": 0.123, "steps": 1, "duration_s": 1.0}
+    sec = _render_collapsible_usage(usage, "MiniMax-M2.7", config=None)
+    assert "(MiniMax-M2.7 — billed)" in sec
     assert "free tier" not in sec
 
 
@@ -1082,7 +1115,10 @@ def test_usage_block_shows_equivalent_provider_cost():
     assert "🔋 AI Usage & Run Details" in sec
     assert "**Est. cost on Claude Sonnet 5**" in sec
     assert "**Actual**: $0.00" in sec
+    # The "free tier" clause must mention the routed model verbatim, not the
+    # stale hardcoded `headroom glm-5.2:cloud` literal.
     assert "free tier" in sec
+    assert "glm-5.2:cloud" in sec
     # Equivalent should be > 0 for non-trivial token counts.
     assert "$0.00" in sec  # the actual line
     # And a non-zero one for the equivalent.