3 models
Ranked by AutomationBench
Reasoning Tools JSON Open weights 30.8

Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work

moonshotai/kimi-k3 2026-07-16 1.04858M context $3/M input $15/M output
62 providers
Reasoning Tools JSON 30.4

High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning

google/gemini-3.7-flash 2026-08-13 1.04858M context $0.75/M input $3.75/M output
23 providers
Reasoning Tools 27.3

Preview Qwen flagship for million-token multimodal reasoning and long-horizon agentic workflows

alibaba/qwen3.8-max-preview 2026-07-19 1M context $2/M input $6/M output
6 providers