From 12ab39167932e1ada5b42855e87fce75a05456f0 Mon Sep 17 00:00:00 2001
From: root <root@island37.com>
Date: Fri, 24 Apr 2026 06:09:34 -0500
Subject: [PATCH] scrum: swap mistral:latest defaults to
 ollama_cloud::gpt-oss:120b
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Three default model lists hardcoded mistral:latest as the fallback
when config.get("model_sets" / "models") returns nothing. Per
feedback_no_mistral.md, mistral 7B has decoder-level JSON malformation
issues (0/5 fill rate on A/B) and is a liability in any path that
depends on structured output from the model.

Swapping to ollama_cloud::gpt-oss:120b (Phase 20 T3 cloud tier)
keeps the defaults reliable for the meta-pipeline orchestrator
(line 9959), the fallback model list for empty Ollama (10084), and
the worker pool default (11835). All three are DEFAULTS — any caller
passing explicit config.model_sets / config.models is unaffected.

Routing works because query_model's "::" provider prefix already
resolves ollama_cloud via commit fa6ccff. Activation requires
OLLAMA_CLOUD_API_KEY or a key saved via the Admin UI; this PR does
not change credential behavior, only the default model list.

Surfaced by lakehouse scrum-master pipeline run 2026-04-24, findings
confirmed by grep verification against the live code.
---
 llm_team_ui.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/llm_team_ui.py b/llm_team_ui.py
index f1bf4db..5fa10fa 100644
--- a/llm_team_ui.py
+++ b/llm_team_ui.py
@@ -9956,7 +9956,7 @@ def _run_meta_pipeline(pipeline_id):
         stages = pipe["stages"] or ["extract", "research", "validate", "synthesize"]
         data_source = pipe["data_source"]
         config = pipe["config"] or {}
-        model_sets = config.get("model_sets", [["qwen2.5:latest"], ["mistral:latest"], ["gemma2:latest"]])
+        model_sets = config.get("model_sets", [["qwen2.5:latest"], ["ollama_cloud::gpt-oss:120b"], ["gemma2:latest"]])
         max_iterations = config.get("max_iterations", len(model_sets))
 
         _meta_status[pipeline_id] = {"stage": 0, "substep": "Gathering data...", "progress": 0, "iteration": 0}
@@ -10081,7 +10081,7 @@ def create_meta_pipeline():
             all_m = [m["name"] for m in resp.json().get("models", []) if m["size"] > 1e9]
             models = [[m] for m in all_m[:4]]
         except Exception:
-            models = [["qwen2.5:latest"], ["mistral:latest"]]
+            models = [["qwen2.5:latest"], ["ollama_cloud::gpt-oss:120b"]]
 
     config = {"model_sets": models, "max_iterations": len(models)}
     with get_db() as conn:
@@ -11832,7 +11832,7 @@ def run_refine(config):
     start = time.time() * 1000
     prompt = config["prompt"]
     orchestrator = config.get("orchestrator", "qwen2.5:latest")
-    workers = config.get("models", ["qwen2.5:latest", "mistral:latest"])
+    workers = config.get("models", ["qwen2.5:latest", "ollama_cloud::gpt-oss:120b"])
     max_stages = config.get("max_stages", 5)
     yield sse({"type": "clear"})
     steps = []