MiniMax M3
MiniMax M3 is an open-weight multimodal AI model designed for coding, agentic workflows, long-context reasoning, and complex automation tasks. The model combines frontier-level coding performance, native multimodal understanding, and a context window of up to 1 million tokens. MiniMax M3 uses MiniMax Sparse Attention to improve long-context efficiency while reducing compute requirements for large-scale inputs. It supports text, image, and video understanding, making it useful for workflows that combine code, documents, visual references, and tool-driven tasks. The model is built for repository-scale reasoning, software engineering, autonomous task execution, tool calling, and multi-step agent workflows. MiniMax M3 helps developers, AI teams, and enterprises build capable agents that can reason across large contexts and work with multimodal information.
Learn more
Letta
Create, deploy, and manage your agents at scale with Letta. Build production applications backed by agent microservices with REST APIs. Letta adds memory to your LLM services to give them advanced reasoning capabilities and transparent long-term memory (powered by MemGPT). We believe that programming agents start with programming memory. Built by the researchers behind MemGPT, introduces self-managed memory for LLMs. Expose the entire sequence of tool calls, reasoning, and decisions that explain agent outputs, right from Letta's Agent Development Environment (ADE). Most systems are built on frameworks that stop at prototyping. Letta' is built by systems engineers for production at scale so the agents you create can increase in utility over time. Interrogate the system, debug your agents, and fine-tune their outputs, all without succumbing to black box services built by Closed AI megacorps.
Learn more
AgentSea
AgentSea is an open source platform designed to build, deploy, and share AI agents with ease. It delivers a collection of libraries and tools for building AI agent apps, favoring the UNIX philosophy of doing one thing well. Tools can be used individually or stacked together into a single agent app, and are compatible with frameworks like LlamaIndex and LangChain. Key components include SurfKit, a Kubernetes-style orchestrator for agents; DeviceBay, offering pluggable devices like file systems and desktops; ToolFuse, a library that wraps scripts, third-party apps, and APIs as Tool implementations; AgentD, a daemon making a Linux desktop OS accessible to bots; AgentDesk, a library for running AgentD-powered VMs; Taskara, for task management; ThreadMem, for building multi-role persistent threads; and MLLM, simplifying communication with multiple LLMs and multimodal LLMs. AgentSea also offers alpha agents like SurfPizza and SurfSlicer, which navigate GUIs using multimodal approaches.
Learn more
LangGraph
Gain precision and control with LangGraph to build agents that reliably handle complex tasks. Build and scale agentic applications with LangGraph Platform.
LangGraph's flexible framework supports diverse control flows – single agent, multi-agent, hierarchical, sequential – and robustly handles realistic, complex scenarios. Ensure reliability with easy-to-add moderation and quality loops that prevent agents from veering off course. Use LangGraph Platform to templatize your cognitive architecture so that tools, prompts, and models are easily configurable with LangGraph Platform Assistants.
With built-in statefulness, LangGraph agents seamlessly collaborate with humans by writing drafts for review and awaiting approval before acting. Easily inspect the agent’s actions and "time-travel" to roll back and take a different action to correct course.
Learn more
