Hermes Agent

🔓 Open

Multi-modal agentic harness with persistent memory, MCP tool orchestration, sub-agent delegation, and autonomous task execution.

💰 Open-source, self-hostable · cli, api, web

72.6Overall Score

Non-Gameable Scoring

Scores are derived from established benchmarks, adjusted for harness-specific performance across four dimensions: Coding, Reasoning, Tool Use, and Autonomy.

Each dimension starts from public benchmark data and applies harness-specific modifiers based on tool integration, context handling, and orchestration quality. The overall score is a weighted composite that penalizes narrow optimization.

Model	Overall	Coding	Reasoning	Tool Use	Autonomy
Claude Opus 4.6	72.6	67.15	80.37	68.28	91.25
Gemini 3.1 Pro	64.6	71.67	77.13	63.89	84.67
Kimi K2.5	62.1	68.34	68.55	43.61	82.9
GPT-5.4	58.1	43.06	59.4	84.67	83.25
Claude Sonnet 4.6	57.3	59.57	49.71	67.09	90.85
MiniMax-M2.7	55.0	53.44	45.7	64.55	83.25
DeepSeek R1	47.9	54.11	45.7	45	66.98
Qwen 3.5	47.4	47.71	46.12	34.78	64.55
Gemini 3 Pro	45.9	39.12	50.61	51.25	66.74
MiMo-V2-Flash	43.0	52.24	42.28	42.16	56.2