Summary

Model pricing for cost calculation is currently hardcoded in deeptutor/agents/research/utils/token_tracker.py. There is no way for users to view or edit prices through the UI, and newly added models like deepseek-v4-pro / deepseek-v4-flash can easily be missed.

Current state

• MODEL_PRICING dict is hardcoded with ~9 models
• Chat UI already displays per-turn cost in dollars
• Settings page has no pricing configuration section
• 2026-05-06: deepseek-v4 entries added to hardcoded table (PR fix: add deepseek-v4-flash and deepseek-v4-pro to model pricing table #447)

Desired

• Add missing deepseek-v4 pricing to hardcoded table (PR fix: add deepseek-v4-flash and deepseek-v4-pro to model pricing table #447)
• Settings page: editable per-model input/output pricing fields
• Backend: store pricing in model catalog or separate config
• Backend: expose pricing via API for the frontend

Why

Users who use non-OpenAI models or custom endpoints see inaccurate cost estimates, and have no way to correct them without editing source code.