From 0e6f5ca5a3f507e6287ab726f45f0f7a17afc72e Mon Sep 17 00:00:00 2001
From: Evandro Fonseca Junior <asemlk54@hotmail.com>
Date: Thu, 26 Mar 2026 17:39:37 -0300
Subject: [PATCH] perf: slim turbo4b runtime profile

---
 crates/inference-core/src/profile.rs | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/crates/inference-core/src/profile.rs b/crates/inference-core/src/profile.rs
index 78d6309..b736d24 100644
--- a/crates/inference-core/src/profile.rs
+++ b/crates/inference-core/src/profile.rs
@@ -15,9 +15,9 @@ impl RuntimeProfile {
         match self {
             Self::Turbo4B => ProfileDefaults {
                 n_gpu_layers: 999,
-                n_ctx: 4096,
-                n_batch: 512,
-                n_ubatch: 128,
+                n_ctx: 2048,
+                n_batch: 256,
+                n_ubatch: 64,
                 flash_attn: true,
                 n_threads: 8,
             },