杨艺韬(yangyitao) 684f41692d fix(ollama): send think param for thinking models
Ollama thinking models (deepseek-r1, qwq, etc.) require an explicit
`think` boolean in the /api/chat request body to control reasoning.
Previously this parameter was never sent, so thinking models always used
their default behavior regardless of the user's thinking-level config.

Now the Ollama stream function forwards the reasoning level as
`think: true` when reasoning is enabled and `think: false` when
reasoning is disabled.

Closes #46680
2026-03-21 02:32:20 +00:00
..
2026-03-15 21:39:49 -07:00
2026-03-17 07:06:25 +00:00
2026-03-17 07:06:25 +00:00