TechnologyAI Agents5 min read

MiniMax Launches MaxClaw Cloud-Hosted OpenClaw Agent on February 25; Clawbot AI Follows with SaaS Version on February 28

MiniMax released MaxClaw on February 25, 2026, as a fully managed cloud deployment of the open-source OpenClaw AI agent framework, running on the MiniMax-M2.5 229-billion-parameter Mixture-of-Experts model. Three days later, Clawbot AI launched its own SaaS version of OpenClaw. Both offerings address the deployment barrier for a framework that accumulated 200,000 GitHub stars in 84 days after launching in November 2025.

A

ObjectWire Technology Desk

MiniMax released MaxClaw on February 25, 2026, as a fully managed cloud mode within the MiniMax Agent platform — enabling one-click deployment of agents built on the open-source OpenClaw framework, powered by the MiniMax-M2.5 229-billion-parameter Mixture-of-Experts model. Three days later, on February 28, Clawbot AI launched its own SaaS cloud platform for OpenClaw, offering managed hosting with automatic AI model selection. Both products target the same deployment friction: OpenClaw is powerful but self-hosted — and the framework had already accumulated 200,000 GitHub stars in 84 days by the time either cloud service launched.

1. OpenClaw: The Open-Source Foundation

OpenClaw is an MIT-licensed personal AI agent framework that runs a local gateway process connecting large language models to real-world tools — browser control, file operations, email, calendar management, and shell commands — across messaging platforms including WhatsApp, Telegram, Slack, and Discord.

OpenClaw Framework — Key Facts

  • •Framework OpenClaw — open-source personal AI agent framework (MIT license)
  • •Launch date November 24, 2025 (originally named Clawdbot, then Moltbot)
  • •GitHub stars 100,000+ in under one week post-launch (late Jan 2026); 200,000+ by February 16, 2026
  • •Time to 200K 84 days from launch
  • •Self-hosted Runs on macOS, Linux, Windows (Docker/WSL2) — local or cloud VM
  • •LLM support Connects any LLM to tools via local gateway process
  • •Platforms WhatsApp, Telegram, Slack, Discord, and others via messaging integrations
  • •Memory Long-term memory stored as Markdown files; proactive scheduling via heartbeat intervals
  • •Tools Browser control, file operations, email, calendar, shell commands, community skills

OpenClaw Growth

200K+

GitHub stars by February 16, 2026

84 days

Time from launch (Nov 24) to 200K stars

<1 week

Time to reach 100,000 stars post late-Jan 2026 surge

Nov 24, 2025

Launch date (originally Clawdbot / Moltbot)

Why the Growth Is Significant

OpenClaw's trajectory — 100,000 GitHub stars in under a week in late January 2026 — mirrors the breakout curves seen with tools like Ollama and LangChain. The framework's appeal is the combination of real messaging platform integration (not just API calls) with persistent memory and proactive scheduling, running locally without any required cloud dependency.

2. MiniMax MaxClaw — Managed Cloud Deployment

MaxClaw operates as a product umbrella covering the OpenClaw framework, managed hosting, and MiniMax-M2.5 as the backend model. The core proposition: deploy a persistent agent connected to messaging apps in under 10 seconds, with no server setup or API key management.

MaxClaw — Key Features

  • •Product MaxClaw — managed cloud mode within MiniMax Agent platform
  • •Launch date February 25, 2026
  • •Access agent.minimax.io / maxclaw.ai
  • •Backend model MiniMax-M2.5 — 229B parameter Mixture-of-Experts, 10B active params/token, 200K context
  • •Setup time One-click deployment in under 10 seconds — no servers or API keys required
  • •Integrations Telegram, WhatsApp (beta), Slack, Discord — 24/7 persistent agents
  • •Templates 10,000+ pre-built Expert templates for task selection
  • •Built-in tools Image generation, video generation, web scraping, file handling, browser control
  • •Model pricing M2.5-Lightning: $0.30/M input tokens · $2.40/M output tokens · 100 tokens/sec

MiniMax-M2.5 — Model Benchmarks

  • •SWE-Bench Verified: 80.2%
  • •Multi-SWE-Bench: 51.3%
  • •BrowseComp: 76.3%
  • •Active params/token: 10B of 229B total (MoE architecture)
  • •Context window: 200,000 tokens
  • •Speed (Lightning): 100 tokens/second

MoE Architecture Note

MiniMax-M2.5's Mixture-of-Experts architecture activates only 10 billion of its 229 billion parameters per token — delivering competitive benchmark performance at inference costs closer to a 10B dense model. The M2.5-Lightning variant targets throughput-sensitive agent workloads at $0.30 per million input tokens and $2.40 per million output tokens at 100 tokens/second.

3. Clawbot AI SaaS — Cloud Platform With Model Selection

Clawbot AI's February 28, 2026 launch introduced a competing managed path for OpenClaw users who want cloud hosting without committing to a single LLM provider. Its differentiator is automatic AI model selection — matching the task at hand to the appropriate model without manual configuration.

Clawbot AI SaaS — Platform Highlights

  • •Product Clawbot AI — hosted SaaS version of OpenClaw
  • •Launch date February 28, 2026
  • •Access clawbot.ai
  • •Runtime Fully managed cloud — no local infrastructure required
  • •Model selection Automatic AI model selection matched to task type

The self-hosted OpenClaw baseline (macOS, Linux, Windows via Docker or WSL2) remains available for teams preferring local or private cloud VM deployment. Both MaxClaw and Clawbot AI add managed layers on top of the same open-source core without forking the upstream project.

4. Ecosystem Context: Two Cloud Wrappers in Three Days

The simultaneous arrival of MaxClaw and Clawbot AI within a 72-hour window reflects a recognizable open-source commercialization pattern: a framework reaches critical adoption mass, and managed service offerings follow rapidly to capture users who want the capability without the infrastructure overhead.

MaxClaw vs. Clawbot AI — Comparison

  • •MaxClaw (MiniMax): Vertically integrated — MiniMax hosts both the agent runtime and the backend LLM (M2.5). 10,000+ Expert templates. Telegram/WhatsApp/Slack/Discord. Launched Feb 25.
  • •Clawbot AI: Bring-your-own-model approach — automatic selection from available models. Focused on managed runtime rather than a proprietary LLM. Launched Feb 28.
  • •Self-hosted OpenClaw: Full control, any LLM, local or cloud VM. Requires Docker or WSL2 setup. MIT license.
  • •Common ground: All three deliver the same OpenClaw agent capabilities — messaging integrations, long-term memory, browser/file/shell tools.

What Comes Next

OpenClaw's star trajectory and the rapid arrival of competing cloud services suggest the agent framework ecosystem is maturing faster than tooling — the next differentiation pressure will likely arrive via enterprise security controls, compliance posture, and multi-agent orchestration rather than basic deployment convenience.

When two cloud wrappers launch for an open-source agent in three days, the barrier isn't the framework anymore — it's choosing which lobster to feed first.