Ce qui s'est passé: The Bitter Lesson of Tool Calling

Researchers compared programmatic tool calling via Python code against JSON-based tool calling across 14 language models on the BFCL v4 benchmark.

Pourquoi ça compte: The Bitter Lesson of Tool Calling

Matters for engineers building LLM agents who need to choose between code-based and JSON-based tool invocation approaches for production systems.

Vigilance: The Bitter Lesson of Tool Calling

Results are specific to BFCL v4 benchmark; real-world performance may vary depending on task complexity, tool diversity, and specific model versions deployed.

Dans l'actualité

The Bitter Lesson of Tool Calling

arXiv cs.AI · Publié le 6 août 2026 · 3 min de lecture

En 30 secondes

Ce qui s'est passé: Researchers compared programmatic tool calling via Python code against JSON-based tool calling across 14 language models on the BFCL v4 benchmark.
Pourquoi ça compte: Matters for engineers building LLM agents who need to choose between code-based and JSON-based tool invocation approaches for production systems.
Vigilance: Results are specific to BFCL v4 benchmark; real-world performance may vary depending on task complexity, tool diversity, and specific model versions deployed.

Écouter ce résumé

Lire l'article complet

agent
llm
language model
tool use
tool calling

The Agent Architect

Un pattern, un compromis, une panne de production racontée. Un brief hebdomadaire court pour ceux qui construisent des systèmes agentiques.

Un email par semaine, désinscription en un clic. Votre adresse ne sert qu'à envoyer le brief.