\begin{table}[htbp]
\centering
\caption{User-selectable model tier. The orchestration service enables adaptive extended
thinking only for models that support it, and applies a uniform per-response generation cap.
The default is a reasoning-augmented model with a very large context window; alternatives let
users trade off capability, latency, and cost per query.}
\label{tab:models}
\begin{threeparttable}
\begin{tabular}{@{}llcc@{}}
\toprule
Role in platform & Model family & Adaptive thinking & Typical use \\
\midrule
Default            & Reasoning-augmented (1M ctx) & Yes & Comprehensive reports \\
High capability    & Flagship reasoning           & Yes & Hardest analyses \\
Balanced           & Mid-tier reasoning           & Yes & Moderate queries \\
Low latency        & Fast tier                    & No  & Simple lookups \\
\bottomrule
\end{tabular}
\begin{tablenotes}[flushleft]
\footnotesize
\item \textit{Notes.} Per-response generation is capped at 64{,}000 tokens. Adaptive thinking
lets the model allocate a variable budget of private reasoning before responding; its summary is
streamed to the client and shown in a collapsible panel. Context-window sizes and per-token
prices are determined by the model provider and abstracted away from the orchestration logic.
\end{tablenotes}
\end{threeparttable}
\end{table}