Upload 4 files

README.md

@@ -133,17 +133,18 @@ See `CHANGELOG_V194.md` for the Model Pre-scan Decision Card UI pass.
 See `CHANGELOG_V195.md` for the Runtime CSS Cleanup with Legacy Safety Net pass.
 
 
-Current release: Agentic Space Factory v198.26.33.
+Current release: Agentic Space Factory v198.26.34.
 
-## v198.26.33 — Model Card Instruction Authority & Terminal Repair Reconciliation
+## v198.26.34 — Model Card Repair Convergence & GLM Default
 
-- Adds `MODEL_CARD_INSTRUCTIONS.json`, extracted from the README/model card, so explicit install commands, loader symbols, runtime notes and model-card deviations become first-class worker context.
-- Promotes explicit model-card requirements into `MODEL_RECIPE.json`, Pi generation prompts, repair task packets and recipe-aware repair packets.
-- Adds model-card alignment and local-requirement sanity guards so generated Spaces cannot silently replace official model-card install commands with missing or non-publishable local artifacts such as invented `.whl` files.
-- Propagates structured `/health` failure classes into recipe-aware repair matching, including SDXL-LoRA `CLIPTextModel.text_model` failures.
-- Strengthens terminal repair reconciliation so `repair_outcome.json` terminal failures cannot leave Active Run or Runs Explorer stuck on `running`.
+- Adds deterministic model-card repair convergence after Pi repair: simple leftovers such as `transformers>=4.51.0` when the card requires `transformers>=4.57` are patched directly and re-checked.
+- Keeps pre-upload repair alive while Pi budget remains, relaunching Pi with direct guard deviations when model-card/template guards still fail.
+- Writes `model_card_repair_convergence_patch.json` with exact requirement changes, removed local artifacts, and README cleanup.
+- Cleans stale local Diffusers wheel artifacts and README “vendored wheel” claims when the final strategy follows an official model-card git dependency.
+- Changes the default Pi assistant model in the Launch Build form and backend fallback to `zai-org/GLM-5.2`.
+- Preserves v198.26.33 model-card instruction authority, local requirement sanity, health-to-repair propagation, terminal repair reconciliation, and v198.26.32 quick links.
 
-See `CHANGELOG_V198_26_33.md`.
+See `CHANGELOG_V198_26_34.md`.
 
 ## v198.26.27 — Pi Repair Execution Guarantee & Model-Family Runtime Recipes
 

app.py

@@ -518,7 +518,7 @@ def _build_live_run_snapshot(
         **({"status": summary_status, "validation_status": summary_status, "result_status": summary_status, "terminal": True, "can_resume": False} if validation_terminal_locked else {}),
     }
 
-    # v198.26.33: repair_outcome.json is a terminal evidence source. A run can
+    # v198.26.34: repair_outcome.json is a terminal evidence source. A run can
     # be left with stale summary/state="running" if the repair path writes the
     # outcome after the last lightweight state flush. Active Run polling should
     # derive the product verdict from this artifact instead of showing endless
@@ -1627,6 +1627,7 @@ Paste a Hugging Face model ID or model-card URL. The worker creates a **private*
             pi_model = gr.Dropdown(
                 label="Pi model",
                 choices=[
+                    "zai-org/GLM-5.2",
                     "Qwen/Qwen3-Coder-Next",
                     "moonshotai/Kimi-K2-Instruct-0905",
                     "zai-org/GLM-4.7",
@@ -1634,7 +1635,7 @@ Paste a Hugging Face model ID or model-card URL. The worker creates a **private*
                     "deepseek-ai/DeepSeek-V3.2",
                     "Qwen/Qwen3-Coder-480B-A35B-Instruct",
                 ],
-                value="Qwen/Qwen3-Coder-Next",
+                value="zai-org/GLM-5.2",
                 allow_custom_value=True,
                 info="Assistant model used by Pi through Hugging Face Inference Providers.",
             )