From 4f30fb3824cbc8d8e56e29a997bd677ed7bea8ad Mon Sep 17 00:00:00 2001
From: Fabio RItzel Borges <38725315+fworks-tech@users.noreply.github.com>
Date: Thu, 17 Sep 2026 15:31:05 -0300
Subject: [PATCH] fix(chat): switch to glm-5.3-flash as mimo-v2.5 is retired

mimo-v2.5 is no longer in the OpenCode Go model list, and mimo-v2.5-free
on zen/v1 rejects server calls with FreeTierError (403). The router
fails inconsistently for mimo-v2.5 (works from some edges, FreeTierError
from Vercel egress). Switches to glm-5.3-flash, verified via direct API
call including tool calling.
---
 src/app/api/chat/route.ts | 4 ++--
 src/lib/abuse/cost.ts     | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/src/app/api/chat/route.ts b/src/app/api/chat/route.ts
index ee9323b..ec667cb 100644
--- a/src/app/api/chat/route.ts
+++ b/src/app/api/chat/route.ts
@@ -40,14 +40,14 @@ export const maxDuration = 60;
 const zen = (sessionId: string) =>
   createOpenAICompatible({
     name: 'zen',
-    baseURL: 'https://opencode.ai/zen/v1',
+    baseURL: 'https://opencode.ai/zen/go/v1',
     headers: {
       Authorization: `Bearer ${process.env.OPENCODE_API_KEY}`,
       'x-opencode-session': sessionId,
     },
   });
 
-const MODEL_ID = 'mimo-v2.5-free';
+const MODEL_ID = 'glm-5.3-flash';
 
 const model = (sessionId: string) => zen(sessionId).chatModel(MODEL_ID);
 
diff --git a/src/lib/abuse/cost.ts b/src/lib/abuse/cost.ts
index f782403..e4112b7 100644
--- a/src/lib/abuse/cost.ts
+++ b/src/lib/abuse/cost.ts
@@ -8,7 +8,7 @@
 
 export const MAX_TOKENS_PER_REQUEST = 4000;
 const MAX_COST_PER_HOUR_USD = 0.5; // ~$0.50/hour budget
-const TOKEN_COST_PER_1M = 0.28; // mimo-v2.5 on OpenCode Go
+const TOKEN_COST_PER_1M = 0.28; // glm-5.3-flash on OpenCode Go (conservative upper bound)
 const USAGE_WINDOW_MS = 60 * 60 * 1000; // 1 hour
 const CLEANUP_INTERVAL_MS = 300_000; // 5 min