diff --git a/cluster/applications/hindsight/values.yaml b/cluster/applications/hindsight/values.yaml index ee780d1..ba5c825 100644 --- a/cluster/applications/hindsight/values.yaml +++ b/cluster/applications/hindsight/values.yaml @@ -49,18 +49,19 @@ existingSecret: hindsight-credentials # ---------------------------------------------------------------------------- # API container environment (explicit env entries; the chart renders this map -# to individual env vars). LLM points at the Nous free-tier inference API -# (OpenAI-compatible). HINDSIGHT_API_LLM_API_KEY is NOT set here — it comes from -# the existingSecret via envFrom (1Password `nous` item). +# to individual env vars). LLM points at the LOCAL astro-orbiter Qwen node +# (OpenAI-compatible) — moved OFF the Nous free tier (solar-pro4 was 404/dead). +# HINDSIGHT_API_LLM_API_KEY is NOT set here — it comes from the existingSecret +# via envFrom (1Password `nous` item); the local node ignores it but the secret +# still satisfies the chart's envFrom contract. # ---------------------------------------------------------------------------- api: env: - HINDSIGHT_API_LLM_BASE_URL: "https://inference-api.nousresearch.com/v1" + HINDSIGHT_API_LLM_BASE_URL: "http://astro-orbiter:8001/v1" HINDSIGHT_API_LLM_PROVIDER: "openai" - # Primary LLM. Timeout raised to 300s (ingress) so a slow agentic reflect - # (~132s on solar-pro4) completes; verified tool-calling + grounded output. - HINDSIGHT_API_LLM_MODEL: "upstage/solar-pro4:free" - # Fallback (documented, NOT deployed): stepfun/step-3.7-flash:free + # Primary LLM: local astro-orbiter Qwen3.8-27B-Q4_K_M (LAN-fast, off Nous free tier). + HINDSIGHT_API_LLM_MODEL: "Qwen3.8-27B-Q4_K_M" + # Fallback (documented, NOT deployed): upstage/solar-pro4:free (Nous) — use if the local node is ever down. # ---------------------------------------------------------------------------- # Ingress via the chart's native template.