Paper adoption · Agent Foundation Model Atlas

Paper–model matrix

Paper	Claude 2	Claude 3.5 Sonnet	Claude 3 Opus	Gemini 1.5 Pro	GLM-5.2	GPT-3.5 Turbo	GPT-4	GPT-4V	GPT-4o	Llama 2 13B Chat	Llama 2 70B Chat	Qwen 7B Chat	Qwen2.5-3B-Instruct	Qwen2.5-7B-Instruct	Qwen3-1.7B	text-davinci-003	Vicuna 13B
AdaPlanner: Adaptive Planning from Feedback with Language Models	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	Policy	—
AgentBench: Evaluating LLMs as Agents	—	—	—	—	—	Baseline	Baseline	—	—	—	Baseline	—	—	—	—	—	Baseline
AgentBoard: An Analytical Evaluation Board of LLM Agents	—	—	—	—	—	Baseline	Baseline	—	—	—	Baseline	Baseline	—	—	—	—	—
AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors	—	—	—	—	—	Actor	Critic	—	—	—	—	—	—	—	—	—	—
AppWorld: A Controllable World of Apps and People for Benchmarking Interactive Agents	—	Baseline	—	—	—	—	Baseline	—	Policy	—	—	—	—	—	—	—	—
CAMEL: Communicative Agents for Mind Exploration of Large Scale Language Model Society	—	—	—	—	—	Actor	—	—	—	—	—	—	—	—	—	—	—
Chameleon: Plug-and-Play Compositional Reasoning with Large Language Models	—	—	—	—	—	Actor	Policy	—	—	—	—	—	—	—	—	—	—
CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing	—	—	—	—	—	Policy	—	—	—	—	Baseline	—	—	—	—	Critic	—
DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines	—	—	—	—	—	Policy	—	—	—	Optimizer	—	—	—	—	—	—	—
ExpeL: LLM Agents Are Experiential Learners	—	—	—	—	—	Reflector	Policy	—	—	—	—	—	—	—	—	—	—
Generative Agents: Interactive Simulacra of Human Behavior	—	—	—	—	—	Policy	—	—	—	—	—	—	—	—	—	—	—
MetaGPT: Meta Programming for Multi-Agent Collaborative Framework	—	—	—	—	—	—	Policy	—	—	—	—	—	—	—	—	—	—
OpenHands: An Open Platform for AI Software Developers as Generalist Agents	—	Actor	—	—	—	—	—	—	Policy	—	—	—	—	—	—	—	—
OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments	—	—	Baseline	Baseline	—	—	Policy	Policy	—	—	—	—	—	—	—	—	—
Reflexion: Language Agents with Verbal Reinforcement Learning	—	—	—	—	—	—	Policy	—	—	—	—	—	—	—	—	—	—
SEED: Self-Evolving On-Policy Distillation for Agentic Reinforcement Learning	—	—	—	—	Analyzer	—	—	—	—	—	—	—	Actor	Actor	Actor	—	—
Self-Refine: Iterative Refinement with Self-Feedback	—	—	—	—	—	Policy	Critic	—	—	—	—	—	—	—	—	—	—
SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering	—	—	—	—	—	—	Policy	—	—	—	—	—	—	—	—	—	—
SWE-bench: Can Language Models Resolve Real-World GitHub Issues?	Baseline	—	—	—	—	Baseline	Baseline	—	—	—	—	—	—	—	—	—	—
Voyager: An Open-Ended Embodied Agent with Large Language Models	—	—	—	—	—	—	Policy	—	—	—	—	—	—	—	—	—	—
WebArena: A Realistic Web Environment for Building Autonomous Agents	—	—	—	—	—	Baseline	Baseline	—	—	—	Baseline	—	—	—	—	Baseline	—

Search the atlas

Self-evolving agent papers

Paper–model matrix