Changes
- Add DeepSeek V4 model support with thinking mode, and set
deepseek-v4-flashas the fast model - Update DeepSeek model list and provider defaults for the new V4 lineup
- Improve earnings output formatting with clearer filing metadata and core figures (revenue, net income, EPS)
Fixes
- Fix memory startup behavior so embedding/index sync failures do not disable keyword search
- Fix embedding calls to time out gracefully instead of hanging during indexing/query embedding
- Fix earnings tool parsing to handle the API’s array response shape correctly