metrics: cost cards show $0.00 for QwenCloud (and DeepSeek/Atlas Cloud/Requesty) — no pricing entries

Author: LearningCircuitCreated Sep 18, 2026Updated Sep 18, 2026

Follow-up from PR #6578 (QwenCloud provider).

metrics/pricing/pricing_fetcher.py has no pricing entries for qwen3.x model ids (the static "qwen": 0.0 row is a local-weights exact-match that never fires for ids like qwen3.8-max). Since QwenCloud is pay-as-you-go, token-cost cards will show $0.00 forever for real spend. The same gap exists on main for DeepSeek, Atlas Cloud, and Requesty models.

Suggested: add current per-token prices for the common qwen3.x ids (and the missing siblings), or render pricing_used: None distinctly from a genuine $0 (e.g. "not priced" instead of $0.00) when no pricing data exists.

Verified in review: cost_calculator converts missing pricing into total_cost: 0.0 with pricing_used: None.

Source: LearningCircuit/local-deep-research