implementation detail · filed under inference & serving
Capped linear reasoning-token length deduction
A token-length deduction is calculated as a capped linear function of reasoning length and effort.
Also called length deduction.
- source
- 1
- model
- 1
- lab adopt it
- 1
- strongest
- core
How sources treat it
One count per evidence span, weakest treatment to strongest.
core 1
Documented in
Evidence
1 span quoted from the sources, strongest treatment first.
the length deduction is rlen(ℓ,b)=−min{Cmax,k(b) ℓ/Lnorm}
coretraining objectivein DeepSeek-V4.1-FlashDeepSeek
Filed alongside
Other methods under inference & serving :: reasoning control.
Configurable reasoning effortChain-of-thought reasoningDefault thinking modeDeployment-time scalar effort controlDisabling reasoning via chat-template configurationInference-time reasoning budget controlInterleaved thinking between tool callsThinking mode selectionCross-turn persistent reasoning historyMaximum thinking effortReasoning parserAlways-on thinking modeclear_thinking chat-template parameterControl-token-enabled thinking modeEffort-dependent exponential token-penalty scheduleGenerate-verify-refine loopMedium-effort reasoning modeParallel-fewest-step samplingQwen3 soft thinking switchTask- and mode-specific sampling parameter recommendationsTest-time compute scalingAdaptive reasoningConfigurable thinking or reasoning modeControllable thinking effort via system message and per-token cost