autoreg-skills-max: initial public release (restored, redacted, hardened)

skills/multi-provider-gateway/references/firepass-models.md

@@ -0,0 +1,31 @@
+# Firepass Model Catalog
+
+Last verified: 2026-07-28
+
+## Endpoint
+- URL: `https://api.firepass.space/v1`
+- Auth: `Bearer fp_live_...`
+
+## Live Models
+
+| Model | Status | Latency |
+|-------|--------|---------|
+| qwen/qwen3.8-max-preview:free | 🟢 LIVE | ~4s |
+
+## Dead Models (all return 403/503)
+
+| Model | HTTP Code |
+|-------|-----------|
+| claude-fable-5 | 403 |
+| deepseek-v4-flash | 503 |
+| gemini-3.6-flash | 403 |
+| gpt-5.4 | 503 |
+| gpt-5.4-mini | 503 |
+| gpt-5.5 | 503 |
+| mimo-v2.5-pro | 503 |
+
+## Notes
+- Firepass has 8 models in catalog but only 1 reliably responds
+- 403 errors suggest the API key may not have access to certain models
+- 503 errors suggest the upstream inference service is down
+- The only working model (qwen/qwen3.8-max-preview:free) is a free-tier preview model