Test whether a model name is a reasoning model

is_reasoning_model() guesses from its name whether a model is a reasoning model: OpenAI's o-series (o1, o3, o4-mini, ...), the GPT-5 and GPT-6 families (such as gpt-6-luna), and any name containing "reasoning". Reasoning models are tuned with reasoning_effort rather than temperature or top_p; gpt-6-luna, for example, accepts temperature and top_p only with reasoning_effort = "none".

Usage

is_reasoning_model(model_name)

Arguments

model_name

A model name, such as "gpt-6-luna" or "o3".

Value

TRUE or FALSE. NULL, NA and empty names give FALSE.

Details

module_parameters() uses this check to decide which parameters to offer for tuning. dsprrr does not change the parameters of calls made with run() for reasoning models.

See also

Other configuration: cache_stats(), clear_cache(), configure_cache(), dsp_configure(), dsprrr_sitrep(), get_default_chat(), with_lm()

Examples

is_reasoning_model("gpt-6-luna")
#> [1] TRUE
is_reasoning_model("o4-mini")
#> [1] TRUE
is_reasoning_model("gpt-4o")
#> [1] FALSE