1h ago
Please support reasoning_effort on deepseek-flash
Currently any request with "reasoning_effort" to deepseek-flash returns 400:
{"error":{"message":"reasoning_effort is supported only for OpenAI
reasoning models."}}
Please accept reasoning_effort for deepseek-flash and apply it.
api.deepseek.com supports the full enum for this model:
none, minimal, low, medium, high, xhigh, ultra, max
(verified none/low/high/max/minimal/medium/xhigh all return 200; invalid
values return the enum above). "none" fully disables reasoning - no
reasoning_content, no reasoning_tokens.
Pending