# OpenClaw

Canonical page: https://agentsky.dev/agents/openclaw

Autonomous agent built for long-running goals.

OpenClaw is an always-on autonomous agent harness designed to pursue long-horizon goals without continuous human direction, staying on-task for hours across complex multi-step workflows. It supports the same broad model set as Hermes and is optimized for tasks that require uninterrupted cloud execution.

## Facts

- **Agent type:** `openclaw`
- **Default model:** GPT-5.6 Sol (`gpt-5.6-sol`)
- **Start in the browser:** https://agentsky.dev/playground/enter?agentType=openclaw&llm=gpt-5.6-sol&source=agent-md
- **API reference:** https://agentsky.dev/docs
- **Billing:** Running sessions are billed by compute time and model tokens; parked agents incur no compute charge.

## Models it runs on

USD per 1M tokens.

| Model | Id | Input | Output |
|---|---|---|---|
| GPT-5.6 Sol (default) | `gpt-5.6-sol` | $3.50 | $21.00 |
| GPT-5.6 Terra | `gpt-5.6-terra` | $1.40 | $8.40 |
| GPT-5.6 Luna | `gpt-5.6-luna` | $0.14 | $0.84 |
| GPT-6 Astra | `gpt-6-astra` | $7.00 | $35.00 |
| GPT-6.1 Sol | `gpt-6.1-sol` | $1.40 | $7.00 |
| GPT-6 Sol | `gpt-6-sol` | $1.40 | $7.00 |
| GPT-6 Luna | `gpt-6-luna` | $0.07 | $0.35 |
| DeepSeek V4 Pro | `deepseek-v4-pro` | $0.924 | $2.772 |
| DeepSeek V4.1 Flash | `deepseek-flash` | $0.21 | $0.84 |
| Gemini 4 Argon | `gemini-4-argon` | $4.40 | $22.00 |
| Gemini 3.5 Flash-Lite | `gemini-3.5-flash-lite` | $0.33 | $2.75 |
| Gemini 3.8 Flash | `gemini-3.8-flash` | $1.65 | $8.25 |
| GLM-5.3 | `glm-5.3` | $1.54 | $4.84 |
| GLM-5.3-Flash | `glm-5.3-flash` | $0.165 | $0.55 |
| Kimi K3.1 Pro | `kimi-k3.1-pro` | $2.20 | $11.00 |
| Kimi K3 | `kimi-k3` | $3.30 | $16.50 |
| Kimi K2.7 Code | `kimi-k2.7-code` | $1.045 | $4.40 |
| Grok 4.6 | `grok-4.6` | $2.20 | $6.60 |
| Claude Opus 5.5 | `claude-opus-5-5` | $2.80 | $14.00 |
| Claude Opus 5 | `claude-opus-5` | $3.50 | $17.50 |
| Claude Fable 5.1 | `claude-fable-5-1` | $7.00 | $35.00 |
| Claude Sonnet 5 | `claude-sonnet-5` | $1.40 | $7.00 |
| Claude Haiku 4.5 | `claude-haiku-4-5` | $0.70 | $3.50 |
| Claude Sonnet 5.5 | `claude-sonnet-5-5` | $1.40 | $7.00 |
| Claude Haiku 5.5 | `claude-haiku-5-5` | $0.07 | $0.35 |
| Grok 4.7 | `grok-4.7` | $2.20 | $6.60 |
| Muse Spark 1.3 | `muse-spark-1.3` | $1.375 | $4.675 |
| Muse Spark 1.2 | `muse-spark-1.2` | $1.375 | $4.675 |
| Muse Spark 1.1 | `muse-spark-1.1` | $1.375 | $4.675 |
| Muse Wonder 1.0 | `muse-wonder-1.0` | $4.40 | $22.00 |
| MiniMax-M3 | `minimax-m3` | $0.33 | $1.32 |
| MiniMax-M3.1-Flash | `minimax-m3.1-flash` | $0.165 | $0.66 |

## What it can do

- **Long-horizon goals:** Pursues multi-hour objectives without continuous human direction.
- **Unattended execution:** Runs autonomously in the cloud — close the laptop immediately after launch.
- **Multi-step planning:** Decomposes ambiguous goals into a concrete action plan and executes it.
- **Any model:** Broad model support — switch provider in the same API call.

## Questions

### How do I start an OpenClaw agent running in the cloud on AgentSky without managing my own server or compute instance?

Sign up for AgentSky and launch an OpenClaw agent via the API or web UI. AgentSky provisions a persistent cloud computer, installs the harness, and keeps it running. Specify the model and goal in the same call. OpenClaw is designed to run unattended — you can close your laptop immediately after starting it.

### Which AI models can I pair with OpenClaw on AgentSky, and which model works best for long autonomous tasks?

OpenClaw supports GPT-5.6 Sol, GPT-5.6 Terra, GPT-5.6 Luna, GPT-6 Astra, GPT-6.1 Sol, GPT-6 Sol, GPT-6 Luna, DeepSeek V4 Pro, DeepSeek V4.1 Flash, Gemini 4 Argon, Gemini 3.5 Flash-Lite, Gemini 3.8 Flash, GLM-5.3, GLM-5.3-Flash, Kimi K3.1 Pro, Kimi K3, Kimi K2.7 Code, Grok 4.6, Claude Opus 5.5, Claude Opus 5, Claude Fable 5.1, Claude Sonnet 5, Claude Haiku 4.5, Claude Sonnet 5.5, Claude Haiku 5.5, Grok 4.7, Muse Spark 1.3, Muse Spark 1.2, Muse Spark 1.1, Muse Wonder 1.0, MiniMax-M3, and MiniMax-M3.1-Flash. For long-horizon tasks, models with larger context windows and stronger instruction-following reduce mid-task drift. Playground coverage for OpenClaw is expanding, so per-model benchmark comparisons are not yet published.

### What does it cost to run an OpenClaw agent for several hours on AgentSky, and is there a free tier or trial available?

Compute is charged for the duration the agent actively runs; parked agents incur no compute charge. Model costs depend on the model selected and token usage per task. Playground coverage for OpenClaw is expanding, so task-level median cost benchmarks are not yet published. Detailed rates are on the pricing page.

### My OpenClaw agent keeps stopping before it finishes a long autonomous task — how do I stop this from happening?

OpenClaw most often stalls on ambiguous decision points where it is uncertain how to proceed without human input. Make the goal and success criteria explicit in the starting prompt — vague goals lead to early stops. Also check whether the task log shows a tool call that failed without a retry; tool errors are a common stop cause. OpenClaw on AgentSky auto-recovers from crashes, so a clean stop is almost always a task-design issue, not a platform failure.

### If an OpenClaw agent is running a multi-hour task and the AgentSky cloud machine it runs on restarts, does it lose everything?

The session workspace persists across ordinary restarts. If the underlying machine is replaced, AgentSky restores the latest durable recovery point, typically preserving the working directory and recent file edits. Work in progress at the exact moment of a hard machine replacement may replay from the last checkpoint. Structuring the goal to produce durable output at each milestone reduces exposure to any replay window.

### How does running OpenClaw on AgentSky compare to self-hosting it or running it locally on my own development machine?

Self-hosting OpenClaw means managing the runtime environment, handling crash recovery, and losing the agent when the host machine goes down. On AgentSky, OpenClaw gets a provisioned persistent cloud computer, managed auto-recovery, and API access with no infrastructure to maintain.