From ad5a72c386dd27d9bf1106dcad003711d798a950 Mon Sep 17 00:00:00 2001
From: Matt Rubens <2600+mrubens@users.noreply.github.com>
Date: Sat, 15 Aug 2026 13:04:36 -0400
Subject: [PATCH 1/5] Proxy Brain reranking through OpenRouter

---
 .docker/gbrain/entrypoint.sh                  | 26 ++++++++++
 .../__tests__/brain-inference.test.ts         | 51 +++++++++++++++++++
 .../api/src/handlers/brain-inference/index.ts | 23 +++++++--
 deploy/compose/docker-compose.prod.yml        |  1 +
 deploy/coolify/docker-compose.yaml            |  1 +
 deploy/railway/template.yaml                  |  1 +
 docker-compose.self-host.yml                  |  1 +
 docker-compose.yml                            |  1 +
 render.yaml                                   |  2 +
 9 files changed, 103 insertions(+), 4 deletions(-)

diff --git a/.docker/gbrain/entrypoint.sh b/.docker/gbrain/entrypoint.sh
index 07e69fbd7..b833d2c6f 100644
--- a/.docker/gbrain/entrypoint.sh
+++ b/.docker/gbrain/entrypoint.sh
@@ -218,6 +218,32 @@ if [ ! -s "$CONFIG_FILE" ]; then
   fi
 fi
 
+# Route gbrain's OpenRouter reranker through the same Roomote credential
+# gateway as embeddings and chat. Do this after initialization so exposing an
+# OpenRouter-compatible endpoint does not change which provider gbrain chooses
+# when it creates the Brain.
+if [ -n "${GBRAIN_RERANKER_MODEL:-}" ]; then
+  case "$GBRAIN_RERANKER_MODEL" in
+    openrouter:*)
+      if [ -z "${OPENROUTER_BASE_URL:-}" ] && [ -n "${OPENAI_BASE_URL:-}" ]; then
+        OPENROUTER_BASE_URL="${OPENAI_BASE_URL%/}/v1"
+        export OPENROUTER_BASE_URL
+      fi
+      if [ -z "${OPENROUTER_API_KEY:-}" ] && [ -n "${OPENAI_API_KEY:-}" ]; then
+        OPENROUTER_API_KEY="$OPENAI_API_KEY"
+        export OPENROUTER_API_KEY
+      fi
+      if [ -z "${OPENROUTER_BASE_URL:-}" ] || [ -z "${OPENROUTER_API_KEY:-}" ]; then
+        echo "[gbrain-entrypoint] WARNING: $GBRAIN_RERANKER_MODEL needs OPENROUTER_BASE_URL and OPENROUTER_API_KEY."
+        echo "[gbrain-entrypoint] WARNING: reranking will remain fail-open until the gateway is configured."
+      fi
+      ;;
+  esac
+
+  gbrain config set search.reranker.model "$GBRAIN_RERANKER_MODEL" >/dev/null
+  echo "[gbrain-entrypoint] reranker: $GBRAIN_RERANKER_MODEL"
+fi
+
 # Adding a key to a brain created without one is a first-class flow rather
 # than an edge case: on hosts whose compose parser ignores `profiles` the
 # service always runs, so a keyless first deploy followed by filling the key
diff --git a/apps/api/src/handlers/brain-inference/__tests__/brain-inference.test.ts b/apps/api/src/handlers/brain-inference/__tests__/brain-inference.test.ts
index f5bcc3a93..9c8fc6f9c 100644
--- a/apps/api/src/handlers/brain-inference/__tests__/brain-inference.test.ts
+++ b/apps/api/src/handlers/brain-inference/__tests__/brain-inference.test.ts
@@ -139,6 +139,57 @@ describe('brain inference gateway', () => {
     });
   });
 
+  it('routes reranking through OpenRouter without exposing its key to gbrain', async () => {
+    const fetchMock = vi.fn(
+      async (_url: string, _init: RequestInit) =>
+        new Response(JSON.stringify({ results: [] }), { status: 200 }),
+    );
+    vi.stubGlobal('fetch', fetchMock);
+
+    const body = {
+      model: 'cohere/rerank-v3.5',
+      query: 'Which result is relevant?',
+      documents: ['relevant', 'unrelated'],
+      top_n: 2,
+    };
+    const response = await post('/v1/rerank', {
+      token: GATEWAY_TOKEN,
+      body,
+    });
+
+    expect(response.status).toBe(200);
+    const [url, init] = fetchMock.mock.calls[0]!;
+    expect(url).toBe('https://openrouter.ai/api/v1/rerank');
+    expect((init.headers as Headers).get('authorization')).toBe(
+      `Bearer ${OPENROUTER.apiKey}`,
+    );
+    expect(JSON.parse(init.body as string)).toEqual(body);
+  });
+
+  it('reports reranking as unavailable when only OpenAI is configured', async () => {
+    mockResolveBrainInferenceProvider.mockResolvedValue({
+      providerId: 'openai',
+      apiKey: 'sk-openai-provider-key',
+    });
+    const fetchMock = vi.fn();
+    vi.stubGlobal('fetch', fetchMock);
+
+    const response = await post('/v1/rerank', {
+      token: GATEWAY_TOKEN,
+      body: {
+        model: 'cohere/rerank-v3.5',
+        query: 'query',
+        documents: ['document'],
+      },
+    });
+
+    expect(response.status).toBe(503);
+    await expect(response.json()).resolves.toMatchObject({
+      error: expect.stringContaining('OpenRouter'),
+    });
+    expect(fetchMock).not.toHaveBeenCalled();
+  });
+
   it('surfaces an unreachable provider as 502 rather than a crash', async () => {
     vi.stubGlobal(
       'fetch',
diff --git a/apps/api/src/handlers/brain-inference/index.ts b/apps/api/src/handlers/brain-inference/index.ts
index d918baf76..4075b411f 100644
--- a/apps/api/src/handlers/brain-inference/index.ts
+++ b/apps/api/src/handlers/brain-inference/index.ts
@@ -18,13 +18,14 @@ import type { Variables } from '../../types';
 const LOG_PREFIX = '[Brain Inference]';
 
 /**
- * The Brain's whole inference surface: embeddings for recall, and one chat
- * path for sourced synthesis and query expansion. Deliberately narrower than
- * the task-sandbox gateway's allowlist, because this credential is a static
- * deployment secret rather than a short-lived run token.
+ * The Brain's whole inference surface: embeddings for recall, reranking for
+ * precision, and chat for sourced synthesis and query expansion. Deliberately
+ * narrower than the task-sandbox gateway's allowlist, because this credential
+ * is a static deployment secret rather than a short-lived run token.
  */
 const BRAIN_ALLOWED_PATHS = new Set([
   '/v1/embeddings',
+  '/v1/rerank',
   '/v1/chat/completions',
   '/v1/responses',
 ]);
@@ -176,6 +177,20 @@ brainInference.post('/*', async (c) => {
     );
   }
 
+  // gbrain's OpenRouter reranker speaks the same authenticated gateway
+  // contract as embeddings and chat, but OpenAI itself has no compatible
+  // rerank endpoint. Fail explicitly instead of forwarding a doomed request
+  // to api.openai.com and obscuring the missing capability as a 404.
+  if (upstreamPath === '/v1/rerank' && resolved.providerId !== 'openrouter') {
+    return c.json(
+      {
+        error:
+          'Brain reranking requires an OpenRouter provider configured in Settings.',
+      },
+      503,
+    );
+  }
+
   const provider = getInferenceGatewayProvider(resolved.providerId);
 
   if (!provider?.authHeader) {
diff --git a/deploy/compose/docker-compose.prod.yml b/deploy/compose/docker-compose.prod.yml
index 9c355f823..cf2f68dc0 100644
--- a/deploy/compose/docker-compose.prod.yml
+++ b/deploy/compose/docker-compose.prod.yml
@@ -534,6 +534,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
+      GBRAIN_RERANKER_MODEL: openrouter:cohere/rerank-v3.5
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/deploy/coolify/docker-compose.yaml b/deploy/coolify/docker-compose.yaml
index 33dee1a02..c0662e6ae 100644
--- a/deploy/coolify/docker-compose.yaml
+++ b/deploy/coolify/docker-compose.yaml
@@ -165,6 +165,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${SERVICE_PASSWORD_64_BRAINGATEWAY}
+      GBRAIN_RERANKER_MODEL: openrouter:cohere/rerank-v3.5
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/deploy/railway/template.yaml b/deploy/railway/template.yaml
index c85b8dd91..d6341c79a 100644
--- a/deploy/railway/template.yaml
+++ b/deploy/railway/template.yaml
@@ -143,6 +143,7 @@ services:
       # goes over the public origin, exactly as TRPC_URL already does.
       OPENAI_BASE_URL: https://${{api.RAILWAY_PUBLIC_DOMAIN}}/api/brain/inference
       OPENAI_API_KEY: ${{api.R_BRAIN_GATEWAY_TOKEN}}
+      GBRAIN_RERANKER_MODEL: openrouter:cohere/rerank-v3.5
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/docker-compose.self-host.yml b/docker-compose.self-host.yml
index 3631c29c8..1ec19d071 100644
--- a/docker-compose.self-host.yml
+++ b/docker-compose.self-host.yml
@@ -297,6 +297,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
+      GBRAIN_RERANKER_MODEL: openrouter:cohere/rerank-v3.5
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/docker-compose.yml b/docker-compose.yml
index f860be001..feae94d2d 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -105,6 +105,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://host.docker.internal:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
+      GBRAIN_RERANKER_MODEL: openrouter:cohere/rerank-v3.5
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/render.yaml b/render.yaml
index 8f12fad5c..ba4851e5b 100644
--- a/render.yaml
+++ b/render.yaml
@@ -162,6 +162,8 @@ services:
           type: web
           name: roomote-api
           envVarKey: R_BRAIN_GATEWAY_TOKEN
+      - key: GBRAIN_RERANKER_MODEL
+        value: openrouter:cohere/rerank-v3.5
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.

From 50016193cc649cfba75cf104fb6e19dbfb1c0c8f Mon Sep 17 00:00:00 2001
From: Matt Rubens <2600+mrubens@users.noreply.github.com>
Date: Sat, 15 Aug 2026 13:13:15 -0400
Subject: [PATCH 2/5] Make the Brain reranker configurable

---
 apps/docs/brain.mdx                    | 25 ++++++++++++++++---------
 deploy/compose/docker-compose.prod.yml |  2 +-
 deploy/coolify/docker-compose.yaml     |  2 +-
 deploy/railway/template.yaml           |  2 +-
 docker-compose.self-host.yml           |  2 +-
 docker-compose.yml                     |  2 +-
 render.yaml                            |  2 +-
 7 files changed, 22 insertions(+), 15 deletions(-)

diff --git a/apps/docs/brain.mdx b/apps/docs/brain.mdx
index c906dee43..57a4533d3 100644
--- a/apps/docs/brain.mdx
+++ b/apps/docs/brain.mdx
@@ -46,9 +46,9 @@ up a provider for tasks, the Brain has what it needs and there is nothing else
 to do. Changing that key later takes effect on the Brain's next request, with
 no redeploy.
 
-OpenRouter and OpenAI are both supported, and the Brain runs the same OpenAI
-models either way, so the choice is about routing and billing rather than
-capability. To bill the Brain separately from task inference, set
+OpenRouter and OpenAI both support the Brain's embedding and synthesis calls,
+but search reranking requires OpenRouter. To bill the Brain separately from
+task inference, set
 `R_BRAIN_OPENROUTER_API_KEY` or `R_BRAIN_OPENAI_API_KEY`; those take
 precedence over the deployment's general provider keys.
 
@@ -82,16 +82,23 @@ in staging is distinguishable from one written against production.
 
 ## Choosing models
 
-Two settings pick the Brain's models:
+Three settings pick the Brain's models:
 
-| Variable                   | What it does      | Written as                     | Changeable            |
-| -------------------------- | ----------------- | ------------------------------ | --------------------- |
-| `R_BRAIN_MODEL`            | Sourced synthesis | your provider's naming         | any time              |
-| `R_BRAIN_EMBEDDING_MODEL`  | Semantic recall   | a plain model id               | before the first boot |
+| Variable                      | What it does      | Written as                        | Changeable            |
+| ----------------------------- | ----------------- | --------------------------------- | --------------------- |
+| `R_BRAIN_MODEL`               | Sourced synthesis | your provider's naming            | any time              |
+| `R_BRAIN_EMBEDDING_MODEL`     | Semantic recall   | a plain model id                  | before the first boot |
+| `GBRAIN_RERANKER_MODEL`       | Search precision  | `openrouter:<provider>/<model-id>` | after a restart       |
 
-Leave both unset and the Brain uses OpenAI's `gpt-5.6-luna` and
+Leave the first two unset and the Brain uses OpenAI's `gpt-5.6-luna` and
 `text-embedding-3-small` through whichever provider you configured.
 
+The reranker defaults to OpenRouter's `voyageai/rerank-2.5-lite`. Set
+`GBRAIN_RERANKER_MODEL` on the gbrain service to choose another model from
+OpenRouter's reranker catalog. Reranking requires an OpenRouter key; with only
+OpenAI configured, gbrain keeps the unreranked results instead of failing the
+search.
+
 The synthesis model is applied by Roomote when it forwards the call and passed
 to the provider as written, so use that provider's naming
 (`openai/gpt-5.6-mini` on OpenRouter, `gpt-5.6-mini` on OpenAI). Changing
diff --git a/deploy/compose/docker-compose.prod.yml b/deploy/compose/docker-compose.prod.yml
index cf2f68dc0..e63ca2e44 100644
--- a/deploy/compose/docker-compose.prod.yml
+++ b/deploy/compose/docker-compose.prod.yml
@@ -534,7 +534,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
-      GBRAIN_RERANKER_MODEL: openrouter:cohere/rerank-v3.5
+      GBRAIN_RERANKER_MODEL: ${GBRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/deploy/coolify/docker-compose.yaml b/deploy/coolify/docker-compose.yaml
index c0662e6ae..649fae577 100644
--- a/deploy/coolify/docker-compose.yaml
+++ b/deploy/coolify/docker-compose.yaml
@@ -165,7 +165,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${SERVICE_PASSWORD_64_BRAINGATEWAY}
-      GBRAIN_RERANKER_MODEL: openrouter:cohere/rerank-v3.5
+      GBRAIN_RERANKER_MODEL: ${GBRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/deploy/railway/template.yaml b/deploy/railway/template.yaml
index d6341c79a..8e40a5d7f 100644
--- a/deploy/railway/template.yaml
+++ b/deploy/railway/template.yaml
@@ -143,7 +143,7 @@ services:
       # goes over the public origin, exactly as TRPC_URL already does.
       OPENAI_BASE_URL: https://${{api.RAILWAY_PUBLIC_DOMAIN}}/api/brain/inference
       OPENAI_API_KEY: ${{api.R_BRAIN_GATEWAY_TOKEN}}
-      GBRAIN_RERANKER_MODEL: openrouter:cohere/rerank-v3.5
+      GBRAIN_RERANKER_MODEL: openrouter:voyageai/rerank-2.5-lite
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/docker-compose.self-host.yml b/docker-compose.self-host.yml
index 1ec19d071..405ef3d11 100644
--- a/docker-compose.self-host.yml
+++ b/docker-compose.self-host.yml
@@ -297,7 +297,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
-      GBRAIN_RERANKER_MODEL: openrouter:cohere/rerank-v3.5
+      GBRAIN_RERANKER_MODEL: ${GBRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/docker-compose.yml b/docker-compose.yml
index feae94d2d..2fe4bd1f4 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -105,7 +105,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://host.docker.internal:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
-      GBRAIN_RERANKER_MODEL: openrouter:cohere/rerank-v3.5
+      GBRAIN_RERANKER_MODEL: ${GBRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/render.yaml b/render.yaml
index ba4851e5b..88eb2075e 100644
--- a/render.yaml
+++ b/render.yaml
@@ -163,7 +163,7 @@ services:
           name: roomote-api
           envVarKey: R_BRAIN_GATEWAY_TOKEN
       - key: GBRAIN_RERANKER_MODEL
-        value: openrouter:cohere/rerank-v3.5
+        value: openrouter:voyageai/rerank-2.5-lite
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.

From f95f1750dbe11c242d6a006968c0966ed5ea4c5a Mon Sep 17 00:00:00 2001
From: Matt Rubens <2600+mrubens@users.noreply.github.com>
Date: Sat, 15 Aug 2026 13:18:44 -0400
Subject: [PATCH 3/5] Preserve versioned reranker base URLs

---
 .docker/gbrain/entrypoint.sh | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/.docker/gbrain/entrypoint.sh b/.docker/gbrain/entrypoint.sh
index b833d2c6f..0a49bcf82 100644
--- a/.docker/gbrain/entrypoint.sh
+++ b/.docker/gbrain/entrypoint.sh
@@ -226,7 +226,11 @@ if [ -n "${GBRAIN_RERANKER_MODEL:-}" ]; then
   case "$GBRAIN_RERANKER_MODEL" in
     openrouter:*)
       if [ -z "${OPENROUTER_BASE_URL:-}" ] && [ -n "${OPENAI_BASE_URL:-}" ]; then
-        OPENROUTER_BASE_URL="${OPENAI_BASE_URL%/}/v1"
+        OPENROUTER_BASE_URL="${OPENAI_BASE_URL%/}"
+        case "$OPENROUTER_BASE_URL" in
+          */v1) ;;
+          *) OPENROUTER_BASE_URL="$OPENROUTER_BASE_URL/v1" ;;
+        esac
         export OPENROUTER_BASE_URL
       fi
       if [ -z "${OPENROUTER_API_KEY:-}" ] && [ -n "${OPENAI_API_KEY:-}" ]; then

From 5d3fac3bfd8abaa5f845b572ceb7d5a0ea9897a3 Mon Sep 17 00:00:00 2001
From: Matt Rubens <2600+mrubens@users.noreply.github.com>
Date: Sat, 15 Aug 2026 13:25:11 -0400
Subject: [PATCH 4/5] Standardize the Brain reranker setting

---
 apps/docs/brain.mdx                    | 4 ++--
 deploy/compose/docker-compose.prod.yml | 3 ++-
 deploy/coolify/README.md               | 7 ++++---
 deploy/coolify/docker-compose.yaml     | 3 ++-
 deploy/railway/README.md               | 7 ++++---
 deploy/railway/template.yaml           | 4 +++-
 deploy/render/README.md                | 7 ++++---
 docker-compose.self-host.yml           | 3 ++-
 docker-compose.yml                     | 2 +-
 render.yaml                            | 7 ++++++-
 10 files changed, 30 insertions(+), 17 deletions(-)

diff --git a/apps/docs/brain.mdx b/apps/docs/brain.mdx
index 57a4533d3..a9771462d 100644
--- a/apps/docs/brain.mdx
+++ b/apps/docs/brain.mdx
@@ -88,13 +88,13 @@ Three settings pick the Brain's models:
 | ----------------------------- | ----------------- | --------------------------------- | --------------------- |
 | `R_BRAIN_MODEL`               | Sourced synthesis | your provider's naming            | any time              |
 | `R_BRAIN_EMBEDDING_MODEL`     | Semantic recall   | a plain model id                  | before the first boot |
-| `GBRAIN_RERANKER_MODEL`       | Search precision  | `openrouter:<provider>/<model-id>` | after a restart       |
+| `R_BRAIN_RERANKER_MODEL`      | Search precision  | `openrouter:<provider>/<model-id>` | after a restart       |
 
 Leave the first two unset and the Brain uses OpenAI's `gpt-5.6-luna` and
 `text-embedding-3-small` through whichever provider you configured.
 
 The reranker defaults to OpenRouter's `voyageai/rerank-2.5-lite`. Set
-`GBRAIN_RERANKER_MODEL` on the gbrain service to choose another model from
+`R_BRAIN_RERANKER_MODEL` to choose another model from
 OpenRouter's reranker catalog. Reranking requires an OpenRouter key; with only
 OpenAI configured, gbrain keeps the unreranked results instead of failing the
 search.
diff --git a/deploy/compose/docker-compose.prod.yml b/deploy/compose/docker-compose.prod.yml
index e63ca2e44..e966b1aa7 100644
--- a/deploy/compose/docker-compose.prod.yml
+++ b/deploy/compose/docker-compose.prod.yml
@@ -24,6 +24,7 @@ x-roomote-base-env: &roomote-base-env
   R_BRAIN_MODEL: ${R_BRAIN_MODEL:-}
   R_BRAIN_EMBEDDING_MODEL: ${R_BRAIN_EMBEDDING_MODEL:-}
   R_BRAIN_EMBEDDING_DIMENSIONS: ${R_BRAIN_EMBEDDING_DIMENSIONS:-}
+  R_BRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
   R_GBRAIN_URL: ${R_GBRAIN_URL:-http://gbrain:8931}
   R_GBRAIN_ADMIN_TOKEN_FILE: /gbrain-data/admin-bootstrap-token
   # Written by the Brain on first boot when no token was supplied, so a stack
@@ -534,7 +535,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
-      GBRAIN_RERANKER_MODEL: ${GBRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
+      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/deploy/coolify/README.md b/deploy/coolify/README.md
index 3986a4b90..cb52f0c6c 100644
--- a/deploy/coolify/README.md
+++ b/deploy/coolify/README.md
@@ -278,9 +278,10 @@ Two operational notes:
   it. `gbrain_data` only holds service configuration and generated bootstrap
   credentials.
 - **Model choice is a variable, not a rebuild.** `R_BRAIN_MODEL` selects the
-  synthesis model and `R_BRAIN_EMBEDDING_MODEL` the embedding model, both in
-  your provider's own naming, both set on the app services. Leave them empty for the
-  defaults. The synthesis model can change at any time; the embedding model
+  synthesis model, `R_BRAIN_EMBEDDING_MODEL` the embedding model, and
+  `R_BRAIN_RERANKER_MODEL` the reranker. Set them on the app services. Leave
+  them empty for the defaults. The synthesis model can change at any time;
+  the reranker changes after a gbrain restart; the embedding model
   sizes the Brain's vector storage when it is first created, so set it (with
   `R_BRAIN_EMBEDDING_DIMENSIONS`) before first boot or not at all. A later
   change is ignored and reported in the Brain's logs rather than silently
diff --git a/deploy/coolify/docker-compose.yaml b/deploy/coolify/docker-compose.yaml
index 649fae577..5d629eba9 100644
--- a/deploy/coolify/docker-compose.yaml
+++ b/deploy/coolify/docker-compose.yaml
@@ -82,6 +82,7 @@ x-roomote-shared-env: &roomote-shared-env
   R_BRAIN_MODEL: ${R_BRAIN_MODEL:-}
   R_BRAIN_EMBEDDING_MODEL: ${R_BRAIN_EMBEDDING_MODEL:-}
   R_BRAIN_EMBEDDING_DIMENSIONS: ${R_BRAIN_EMBEDDING_DIMENSIONS:-}
+  R_BRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
   R_GBRAIN_URL: http://gbrain:8931
   # Roomote uses this only to register its own scoped clients against the
   # Brain. Coolify's SERVICE_PASSWORD_64_* generates an
@@ -165,7 +166,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${SERVICE_PASSWORD_64_BRAINGATEWAY}
-      GBRAIN_RERANKER_MODEL: ${GBRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
+      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/deploy/railway/README.md b/deploy/railway/README.md
index d081d726f..c51f65d5d 100644
--- a/deploy/railway/README.md
+++ b/deploy/railway/README.md
@@ -429,9 +429,10 @@ Two operational notes:
   re-registers its clients automatically when the Brain no longer recognizes
   them — but the deployment starts cold until that finishes.
 - **Model choice is a variable, not a rebuild.** `R_BRAIN_MODEL` selects the
-  synthesis model and `R_BRAIN_EMBEDDING_MODEL` the embedding model, both in
-  your provider's own naming, both set on **api**. Leave them empty for the
-  defaults. The synthesis model can change at any time; the embedding model
+  synthesis model, `R_BRAIN_EMBEDDING_MODEL` the embedding model, and
+  `R_BRAIN_RERANKER_MODEL` the reranker, all set on **api**. Leave them empty
+  for the defaults. The synthesis model can change at any time; the reranker
+  changes after a gbrain restart; the embedding model
   sizes the Brain's vector storage when it is first created, so set it (with
   `R_BRAIN_EMBEDDING_DIMENSIONS`) before first boot or not at all. A later
   change is ignored and reported in the Brain's logs rather than silently
diff --git a/deploy/railway/template.yaml b/deploy/railway/template.yaml
index 8e40a5d7f..278c0a22d 100644
--- a/deploy/railway/template.yaml
+++ b/deploy/railway/template.yaml
@@ -143,7 +143,7 @@ services:
       # goes over the public origin, exactly as TRPC_URL already does.
       OPENAI_BASE_URL: https://${{api.RAILWAY_PUBLIC_DOMAIN}}/api/brain/inference
       OPENAI_API_KEY: ${{api.R_BRAIN_GATEWAY_TOKEN}}
-      GBRAIN_RERANKER_MODEL: openrouter:voyageai/rerank-2.5-lite
+      GBRAIN_RERANKER_MODEL: ${{api.R_BRAIN_RERANKER_MODEL}}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
@@ -213,6 +213,7 @@ services:
       R_BRAIN_MODEL: ''
       R_BRAIN_EMBEDDING_MODEL: ''
       R_BRAIN_EMBEDDING_DIMENSIONS: ''
+      R_BRAIN_RERANKER_MODEL: openrouter:voyageai/rerank-2.5-lite
       # Railway's private network is IPv6-only and never leaves the project,
       # so the Brain is unreachable from the internet by construction.
       R_GBRAIN_URL: http://${{gbrain.RAILWAY_PRIVATE_DOMAIN}}:8931
@@ -260,6 +261,7 @@ services:
       R_BRAIN_GATEWAY_TOKEN: ${{api.R_BRAIN_GATEWAY_TOKEN}}
       R_BRAIN_MODEL: ${{api.R_BRAIN_MODEL}}
       R_BRAIN_EMBEDDING_MODEL: ${{api.R_BRAIN_EMBEDDING_MODEL}}
+      R_BRAIN_RERANKER_MODEL: ${{api.R_BRAIN_RERANKER_MODEL}}
       R_GBRAIN_URL: ${{api.R_GBRAIN_URL}}
       R_GBRAIN_ADMIN_TOKEN: ${{api.R_GBRAIN_ADMIN_TOKEN}}
       R_APP_ENV: ${{api.R_APP_ENV}}
diff --git a/deploy/render/README.md b/deploy/render/README.md
index bd480330a..936b9295d 100644
--- a/deploy/render/README.md
+++ b/deploy/render/README.md
@@ -392,9 +392,10 @@ Two operational notes:
   longer recognizes them — but the deployment starts cold until that
   finishes.
 - **Model choice is a variable, not a rebuild.** `R_BRAIN_MODEL` selects the
-  synthesis model and `R_BRAIN_EMBEDDING_MODEL` the embedding model, both in
-  your provider's own naming, both set on the app services. Leave them empty for the
-  defaults. The synthesis model can change at any time; the embedding model
+  synthesis model, `R_BRAIN_EMBEDDING_MODEL` the embedding model, and
+  `R_BRAIN_RERANKER_MODEL` the reranker. Set them on the api service. Leave
+  them empty for the defaults. The synthesis model can change at any time;
+  the reranker changes after a gbrain restart; the embedding model
   sizes the Brain's vector storage when it is first created, so set it (with
   `R_BRAIN_EMBEDDING_DIMENSIONS`) before first boot or not at all. A later
   change is ignored and reported in the Brain's logs rather than silently
diff --git a/docker-compose.self-host.yml b/docker-compose.self-host.yml
index 405ef3d11..1b404cae3 100644
--- a/docker-compose.self-host.yml
+++ b/docker-compose.self-host.yml
@@ -36,6 +36,7 @@ x-roomote-env: &roomote-env
   R_BRAIN_MODEL: ${R_BRAIN_MODEL:-}
   R_BRAIN_EMBEDDING_MODEL: ${R_BRAIN_EMBEDDING_MODEL:-}
   R_BRAIN_EMBEDDING_DIMENSIONS: ${R_BRAIN_EMBEDDING_DIMENSIONS:-}
+  R_BRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
   R_GBRAIN_URL: ${R_GBRAIN_URL:-http://gbrain:8931}
   # Roomote reads the brain's bootstrap token once to register its own
   # scoped clients; api and bullmq mount the brain volume read-only.
@@ -297,7 +298,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
-      GBRAIN_RERANKER_MODEL: ${GBRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
+      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/docker-compose.yml b/docker-compose.yml
index 2fe4bd1f4..3b7b679a9 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -105,7 +105,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://host.docker.internal:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
-      GBRAIN_RERANKER_MODEL: ${GBRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
+      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/render.yaml b/render.yaml
index 88eb2075e..3be359658 100644
--- a/render.yaml
+++ b/render.yaml
@@ -163,7 +163,10 @@ services:
           name: roomote-api
           envVarKey: R_BRAIN_GATEWAY_TOKEN
       - key: GBRAIN_RERANKER_MODEL
-        value: openrouter:voyageai/rerank-2.5-lite
+        fromService:
+          type: web
+          name: roomote-api
+          envVarKey: R_BRAIN_RERANKER_MODEL
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
@@ -250,6 +253,8 @@ services:
         sync: false
       - key: R_BRAIN_EMBEDDING_MODEL
         sync: false
+      - key: R_BRAIN_RERANKER_MODEL
+        value: openrouter:voyageai/rerank-2.5-lite
       - key: ROOMOTE_GBRAIN_HOSTPORT
         fromService:
           type: pserv

From 5944fba77f8ebbd5b2cf9385c9ea15ee48e8d8e6 Mon Sep 17 00:00:00 2001
From: Matt Rubens <2600+mrubens@users.noreply.github.com>
Date: Sat, 15 Aug 2026 13:32:49 -0400
Subject: [PATCH 5/5] Restore the default Brain reranker when unset

---
 .docker/gbrain/entrypoint.sh           | 50 +++++++++++++-------------
 deploy/compose/docker-compose.prod.yml |  4 +--
 deploy/coolify/docker-compose.yaml     |  4 +--
 deploy/railway/template.yaml           |  2 +-
 docker-compose.self-host.yml           |  4 +--
 docker-compose.yml                     |  2 +-
 render.yaml                            |  2 +-
 7 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/.docker/gbrain/entrypoint.sh b/.docker/gbrain/entrypoint.sh
index 0a49bcf82..e0ecba2e7 100644
--- a/.docker/gbrain/entrypoint.sh
+++ b/.docker/gbrain/entrypoint.sh
@@ -221,32 +221,32 @@ fi
 # Route gbrain's OpenRouter reranker through the same Roomote credential
 # gateway as embeddings and chat. Do this after initialization so exposing an
 # OpenRouter-compatible endpoint does not change which provider gbrain chooses
-# when it creates the Brain.
-if [ -n "${GBRAIN_RERANKER_MODEL:-}" ]; then
-  case "$GBRAIN_RERANKER_MODEL" in
-    openrouter:*)
-      if [ -z "${OPENROUTER_BASE_URL:-}" ] && [ -n "${OPENAI_BASE_URL:-}" ]; then
-        OPENROUTER_BASE_URL="${OPENAI_BASE_URL%/}"
-        case "$OPENROUTER_BASE_URL" in
-          */v1) ;;
-          *) OPENROUTER_BASE_URL="$OPENROUTER_BASE_URL/v1" ;;
-        esac
-        export OPENROUTER_BASE_URL
-      fi
-      if [ -z "${OPENROUTER_API_KEY:-}" ] && [ -n "${OPENAI_API_KEY:-}" ]; then
-        OPENROUTER_API_KEY="$OPENAI_API_KEY"
-        export OPENROUTER_API_KEY
-      fi
-      if [ -z "${OPENROUTER_BASE_URL:-}" ] || [ -z "${OPENROUTER_API_KEY:-}" ]; then
-        echo "[gbrain-entrypoint] WARNING: $GBRAIN_RERANKER_MODEL needs OPENROUTER_BASE_URL and OPENROUTER_API_KEY."
-        echo "[gbrain-entrypoint] WARNING: reranking will remain fail-open until the gateway is configured."
-      fi
-      ;;
-  esac
+# when it creates the Brain. An empty forwarded setting restores the default,
+# including after a deployment previously selected another reranker.
+GBRAIN_RERANKER_MODEL="${GBRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}"
+case "$GBRAIN_RERANKER_MODEL" in
+  openrouter:*)
+    if [ -z "${OPENROUTER_BASE_URL:-}" ] && [ -n "${OPENAI_BASE_URL:-}" ]; then
+      OPENROUTER_BASE_URL="${OPENAI_BASE_URL%/}"
+      case "$OPENROUTER_BASE_URL" in
+        */v1) ;;
+        *) OPENROUTER_BASE_URL="$OPENROUTER_BASE_URL/v1" ;;
+      esac
+      export OPENROUTER_BASE_URL
+    fi
+    if [ -z "${OPENROUTER_API_KEY:-}" ] && [ -n "${OPENAI_API_KEY:-}" ]; then
+      OPENROUTER_API_KEY="$OPENAI_API_KEY"
+      export OPENROUTER_API_KEY
+    fi
+    if [ -z "${OPENROUTER_BASE_URL:-}" ] || [ -z "${OPENROUTER_API_KEY:-}" ]; then
+      echo "[gbrain-entrypoint] WARNING: $GBRAIN_RERANKER_MODEL needs OPENROUTER_BASE_URL and OPENROUTER_API_KEY."
+      echo "[gbrain-entrypoint] WARNING: reranking will remain fail-open until the gateway is configured."
+    fi
+    ;;
+esac
 
-  gbrain config set search.reranker.model "$GBRAIN_RERANKER_MODEL" >/dev/null
-  echo "[gbrain-entrypoint] reranker: $GBRAIN_RERANKER_MODEL"
-fi
+gbrain config set search.reranker.model "$GBRAIN_RERANKER_MODEL" >/dev/null
+echo "[gbrain-entrypoint] reranker: $GBRAIN_RERANKER_MODEL"
 
 # Adding a key to a brain created without one is a first-class flow rather
 # than an edge case: on hosts whose compose parser ignores `profiles` the
diff --git a/deploy/compose/docker-compose.prod.yml b/deploy/compose/docker-compose.prod.yml
index e966b1aa7..cabdcba9f 100644
--- a/deploy/compose/docker-compose.prod.yml
+++ b/deploy/compose/docker-compose.prod.yml
@@ -24,7 +24,7 @@ x-roomote-base-env: &roomote-base-env
   R_BRAIN_MODEL: ${R_BRAIN_MODEL:-}
   R_BRAIN_EMBEDDING_MODEL: ${R_BRAIN_EMBEDDING_MODEL:-}
   R_BRAIN_EMBEDDING_DIMENSIONS: ${R_BRAIN_EMBEDDING_DIMENSIONS:-}
-  R_BRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
+  R_BRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-}
   R_GBRAIN_URL: ${R_GBRAIN_URL:-http://gbrain:8931}
   R_GBRAIN_ADMIN_TOKEN_FILE: /gbrain-data/admin-bootstrap-token
   # Written by the Brain on first boot when no token was supplied, so a stack
@@ -535,7 +535,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
-      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
+      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/deploy/coolify/docker-compose.yaml b/deploy/coolify/docker-compose.yaml
index 5d629eba9..75592f21b 100644
--- a/deploy/coolify/docker-compose.yaml
+++ b/deploy/coolify/docker-compose.yaml
@@ -82,7 +82,7 @@ x-roomote-shared-env: &roomote-shared-env
   R_BRAIN_MODEL: ${R_BRAIN_MODEL:-}
   R_BRAIN_EMBEDDING_MODEL: ${R_BRAIN_EMBEDDING_MODEL:-}
   R_BRAIN_EMBEDDING_DIMENSIONS: ${R_BRAIN_EMBEDDING_DIMENSIONS:-}
-  R_BRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
+  R_BRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-}
   R_GBRAIN_URL: http://gbrain:8931
   # Roomote uses this only to register its own scoped clients against the
   # Brain. Coolify's SERVICE_PASSWORD_64_* generates an
@@ -166,7 +166,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${SERVICE_PASSWORD_64_BRAINGATEWAY}
-      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
+      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/deploy/railway/template.yaml b/deploy/railway/template.yaml
index 278c0a22d..f1a54afc3 100644
--- a/deploy/railway/template.yaml
+++ b/deploy/railway/template.yaml
@@ -213,7 +213,7 @@ services:
       R_BRAIN_MODEL: ''
       R_BRAIN_EMBEDDING_MODEL: ''
       R_BRAIN_EMBEDDING_DIMENSIONS: ''
-      R_BRAIN_RERANKER_MODEL: openrouter:voyageai/rerank-2.5-lite
+      R_BRAIN_RERANKER_MODEL: ''
       # Railway's private network is IPv6-only and never leaves the project,
       # so the Brain is unreachable from the internet by construction.
       R_GBRAIN_URL: http://${{gbrain.RAILWAY_PRIVATE_DOMAIN}}:8931
diff --git a/docker-compose.self-host.yml b/docker-compose.self-host.yml
index 1b404cae3..59691603d 100644
--- a/docker-compose.self-host.yml
+++ b/docker-compose.self-host.yml
@@ -36,7 +36,7 @@ x-roomote-env: &roomote-env
   R_BRAIN_MODEL: ${R_BRAIN_MODEL:-}
   R_BRAIN_EMBEDDING_MODEL: ${R_BRAIN_EMBEDDING_MODEL:-}
   R_BRAIN_EMBEDDING_DIMENSIONS: ${R_BRAIN_EMBEDDING_DIMENSIONS:-}
-  R_BRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
+  R_BRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-}
   R_GBRAIN_URL: ${R_GBRAIN_URL:-http://gbrain:8931}
   # Roomote reads the brain's bootstrap token once to register its own
   # scoped clients; api and bullmq mount the brain volume read-only.
@@ -298,7 +298,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://api:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
-      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
+      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/docker-compose.yml b/docker-compose.yml
index 3b7b679a9..74509c7cb 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -105,7 +105,7 @@ services:
       # provider key below instead makes the Brain call the provider directly.
       OPENAI_BASE_URL: ${GBRAIN_OPENAI_BASE_URL:-http://host.docker.internal:3001/api/brain/inference}
       OPENAI_API_KEY: ${R_BRAIN_GATEWAY_TOKEN:-}
-      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-openrouter:voyageai/rerank-2.5-lite}
+      GBRAIN_RERANKER_MODEL: ${R_BRAIN_RERANKER_MODEL:-}
       # Create-time, and it has to be the container that receives these: the
       # embedding model and its width are decided when the Brain is created,
       # and the gateway must never substitute a different one afterwards.
diff --git a/render.yaml b/render.yaml
index 3be359658..ff7ab12db 100644
--- a/render.yaml
+++ b/render.yaml
@@ -254,7 +254,7 @@ services:
       - key: R_BRAIN_EMBEDDING_MODEL
         sync: false
       - key: R_BRAIN_RERANKER_MODEL
-        value: openrouter:voyageai/rerank-2.5-lite
+        value: ''
       - key: ROOMOTE_GBRAIN_HOSTPORT
         fromService:
           type: pserv