TRIP-Bench: A Benchmark for Long-Horizon Interactive Agents in Real-World Scenarios

As LLM-based agents are deployed in increasingly complex real-world settings, existing benchmarks underrepresent key challenges such as enforcing global constraints, coordinating multi-tool reasoning, and adapting to evolving user behavior over long, multi-turn interactions. To bridge this gap, we introduce \textbf{TRIP-Bench}, a long-horizon benchmark grounded in realistic travel-planning scenarios. TRIP-Bench leverages real-world data, offers 18 curated tools and 40+ travel requirements, and supports automated evaluation. It includes splits of varying difficulty; the hard split emphasizes long and ambiguous interactions, style shifts, feasibility changes, and iterative version revision. Dialogues span up to 15 user turns, can involve 150+ tool calls, and may exceed 200k tokens of context. Experiments show that even advanced models achieve at most 50\% success on the easy split, with performance dropping below 10\% on hard subsets. We further propose GTPO, an online multi-turn reinforcement learning method with specialized reward normalization and reward differencing. Applied to Qwen2.5-32B-Instruct, GTPO improves constraint satisfaction and interaction robustness, outperforming Gemini-3-Pro in our evaluation. We expect TRIP-Bench to advance practical long-horizon interactive agents, and GTPO to provide an effective online RL recipe for robust long-horizon training.

AI agents are expected to help users complete complex tasks, but real conversations are often long, changing, and full of constraints. For example, in travel planning, users may revise budgets, add preferences, or ask for new versions of a plan. We introduce TRIP-Bench, a realistic travel-planning benchmark that tests whether AI agents can use tools, remember user needs, follow rules, and revise plans across many turns. Experiments show that even strong models still struggle on these long-horizon tasks. We also propose GTPO, a reinforcement learning method that improves agents’ ability to follow constraints and adapt during interaction. This work helps evaluate and train more reliable AI agents for complex real-world tasks.