Edit Issue - .issue.db

Title *

Description

ROOT CAUSE: engine/tools/llm/usage.py PRICING dict has no 'ollama' provider key. The agent route (config.ini [llm.routes] agent = ollama/glm-5:cloud, ollama/qwen3.5:397b-cloud, ollama/minimax-m3:cloud) runs on Ollama CLOUD ([ollama] base_url=https://ollama.com — PAID, not local), so estimate_cost_usd('ollama', 'glm-5:cloud', ...) hits PRICING.get('ollama', []) = [] and returns 0.0 for every call. Confirmed: 14636/14636 llm_call agent_events have estimated_cost_usd=0 despite non-zero tokens (e.g. 19393). IMPACT: (1) cost reporting (trace tab, token_usage, run_budget.usd_spent) is always $0 for the primary models; (2) the [llm.budget] max_usd_per_run cost rail can NEVER trip for these models (usd_spent stays 0) — only the token rail (max_tokens_per_run) works. Not a calc bug: unknown-model->$0 is by design (graceful). FIX: add an 'ollama' entry to PRICING with real per-token prices for glm-5:cloud / qwen3.5:397b-cloud / minimax-m3:cloud (Ollama Cloud pricing — authoritative numbers needed, do NOT invent). SECONDARY: estimate_cost_usd silently returns 0 for unpriced models, indistinguishable from genuinely-free — consider a one-time WARN log per unpriced provider/model so operators know cost tracking has a gap.

Due Date (YYYY-MM-DD)

Tags (comma separated)

Related Issues (IDs)

Enter IDs of issues related to this one. They will be linked as 'related'.

Edit Issue #808