[{"model":"Gemini-3-Flash-Preview","agent":"ReAct (Prompt)","agent_url":"https://github.com/IBM/vakra","date":"Mar 19, 2026","scores":{"api_chaining":35.1,"tool_selection":60.3,"multihop":38.43,"multiturn":18.43},"overall":38.06,"rank":1},{"model":"GPT-OSS-120B","agent":"ReAct (Prompt)","agent_url":"https://github.com/IBM/vakra","date":"Mar 24, 2026","scores":{"api_chaining":40.2,"tool_selection":50.46,"multihop":29.4,"multiturn":16.68},"overall":34.19,"rank":2},{"model":"Claude-Sonnet-4.5","agent":"ReAct (Prompt)","agent_url":"https://github.com/IBM/vakra","date":"Mar 19, 2026","scores":{"api_chaining":34.0,"tool_selection":49.5,"multihop":30.08,"multiturn":17.61},"overall":32.8,"rank":3},{"model":"LLAMA-405B","agent":"ReAct (Prompt)","agent_url":"https://github.com/IBM/vakra","date":"Mar 19, 2026","scores":{"api_chaining":34.56,"tool_selection":54.35,"multihop":27.59,"multiturn":13.18},"overall":32.42,"rank":4},{"model":"Granite-4.0-h-small-32B","agent":"ReAct (Prompt)","agent_url":"https://github.com/IBM/vakra","date":"Mar 19, 2026","scores":{"api_chaining":23.3,"tool_selection":50.0,"multihop":25.8,"multiturn":15.05},"overall":28.54,"rank":5},{"model":"Qwen3.6-35B-A3B","agent":"Juris","agent_url":"https://research.clerktree.com/juris","date":"May 21, 2026","scores":{"api_chaining":0.0,"tool_selection":59.42,"multihop":0.0,"multiturn":0.0},"overall":14.86,"rank":6}]