Files
learning-garden/sources/2026-04-20-model-check.md

58 lines
1.9 KiB
Markdown
Raw Permalink Normal View History

---
pageType: source
id: source.2026-04-20-model-check
title: 2026-04-20-model-check
sourceType: local-file
sourcePath: /home/topher/.openclaw/workspace-crash-bot/memory/2026-04-20-model-check.md
ingestedAt: 2026-05-03T01:02:46.917Z
updatedAt: 2026-05-03T01:02:46.917Z
status: active
growth: seed
---
# 2026-04-20-model-check
## Source
- Type: `local-file`
- Path: `/home/topher/.openclaw/workspace-crash-bot/memory/2026-04-20-model-check.md`
- Bytes: 1225
- Updated: 2026-05-03T01:02:46.917Z
## Content
```text
# 2026-04-20 — Model Check Session
## What Happened
-topher noticed he was on minimax-m2.7 instead of the configured GLM-5.1:cloud.
## Investigation
- Both crash-bot and psb-thinking have identical model config: `glm-5.1:cloud` primary, `minimax-m2.7` fallback
- Earlier curl tests showed timeouts to both cloud endpoints — transient outage
- Later curl test confirmed GLM-5.1:cloud IS alive and responding on `https://ollama.com/v1/chat/completions`
- minimax also works but was the fallback during the outage window
- Confirmed: `api.ollama.com` redirects to `ollama.com` — endpoint shifted
## Key Learnings
- GLM and minimax are both working cloud models
- -topher likes to experiment with new models as they drop — GLM is the lab rat, minimax is the safety net
- `/new` spins up a fresh session that reads current config (GLM primary)
- Current session was probably locked to minimax during GLM's transient outage window
## Model Status
- crash-bot: ollama/glm-5.1:cloud (primary) / ollama/minimax-m2.7 (fallback) ✅
- psb-thinking: same config ✅
- Local Ollama: only embedding models (all-minilm, nomic-embed-text) — no local chat models
## No config changes made — everything is working as intended
```
## Notes
<!-- openclaw:human:start -->
<!-- openclaw:human:end -->
## Related
<!-- openclaw:wiki:related:start -->
- No related pages yet.
<!-- openclaw:wiki:related:end -->