lusty swan
Regarding the "Thinking" capability of Claude 3.7 Sonnet, does it support parameter customization?
I noticed on Openrouter that there are currently two models available:
anthropic/claude-3.7-sonnet
anthropic/claude-3.7-sonnet:thinking
In the Anthropic documentation, the definition of "Extended Thinking" is as follows:
'''
"thinking": {
"type": "enabled",
"budget_tokens": 32000
}
''''
Since these are now split into two distinct model versions, does this imply that thinking type: enabled is already integrated by default into claude-3.7-sonnet:thinking? Or do we still need to define it manually? If customization is required, how can we specify the budget_tokens parameter for thinking?
glacial mesa