Skip to content

YC (S26) | Open Computer History | Continuously record your company computer work, map your workflows, help you find work worth automating, and power your agents' context

feat(ai-gateway): switch auto waterfall primary from gemini-3.5-flash to glm-5

Louis Beaumontd2ec1e0e9ab5
diff --git a/packages/ai-gateway/src/handlers/chat.ts b/packages/ai-gateway/src/handlers/chat.ts
index acc0c8430..b19dbdca1 100644
--- a/packages/ai-gateway/src/handlers/chat.ts
+++ b/packages/ai-gateway/src/handlers/chat.ts
@@ -7,11 +7,10 @@ import { addCorsHeaders } from '../utils/cors';
import { logModelOutcome } from '../services/model-health';
import { captureException } from '@sentry/cloudflare';
-// Auto model waterfall — quality first, then fall back to open-weight Vertex
-// MaaS picks (free for users, near-zero GCP burn) and a cheap Gemini safety
-// net at the tail.
+// Auto model waterfall — open-weight Vertex MaaS picks ordered by quality
+// (free for users, low GCP burn) with a cheap Gemini safety net at the tail.
const AUTO_WATERFALL = [
- 'gemini-3.5-flash', // new flagship flash — beats pro tier on agent benchmarks
+ 'glm-5', // closest Vertex match to gemini-3.5-flash (AA index 50 vs 55) at ~1/3 the output cost
'kimi-k2.5',
'deepseek-v3.2',
'glm-4.7',