OpenCompress Plugin for OpenClaw

Compress every LLM call automatically — keep your existing provider, same models, same quality, 40-70% cheaper.

How it works

Your Agent → OpenCompress (compress) → Your LLM Provider (OpenAI/Anthropic/OpenRouter/Google)

You already pay for an LLM provider. OpenCompress adds a compression layer on top — your prompts are compressed through a 5-layer pipeline before reaching your provider. You pay your provider at their normal rates, and we charge only 20% of what you save.

• 53% average input token reduction
• 62% latency improvement
• 96% quality preservation (SQuALITY benchmark)

Install

openclaw plugins install @opencompress/opencompress

Setup

1. Connect your LLM key

After installing, run onboard and connect your existing provider key:

openclaw onboard opencompress

The wizard auto-provisions your account ($1.00 free credit) and asks for your upstream LLM key. Supported providers:

| Key prefix | Provider | |---|---| | sk-proj- or sk- | OpenAI | | sk-ant- | Anthropic | | sk-or- | OpenRouter | | AIza... | Google AI |

Once connected, every LLM call is compressed automatically — you pay your provider directly, we only charge the compression fee.

Don't have an LLM key? No problem — we can route through OpenRouter for you. Just skip the key step during onboard.

2. Use it

Switch to the OpenCompress provider:

/model opencompress/gpt-4o-mini

That's it. Same model IDs as your current provider — no config changes needed.

3. Connect or switch your key anytime

/compress-byok sk-proj-your-openai-key    # Connect OpenAI
/compress-byok sk-ant-your-anthropic-key   # Connect Anthropic
/compress-byok sk-or-your-openrouter-key   # Connect OpenRouter
/compress-byok off                         # Switch back to router mode

Commands

| Command | Description | |---------|-------------| | /compress-stats | Show compression savings (calls, tokens saved, cost saved) | | /compress-byok <key> | Connect or switch your LLM provider key | | /compress-byok off | Disconnect your key (switch to router fallback) |

Supported models (20)

Works with all major providers — use whichever models you already use:

| Provider | Models | |----------|--------| | OpenAI | gpt-4o, gpt-4o-mini, gpt-4.1, gpt-4.1-mini, gpt-4.1-nano, o3, o4-mini | | Anthropic | claude-sonnet-4-6, claude-opus-4-6, claude-haiku-4-5-20251001 | | Google | gemini-2.5-pro, gemini-2.5-flash, google/gemini-2.5-pro-preview | | DeepSeek | deepseek/deepseek-chat-v3-0324, deepseek/deepseek-reasoner | | Meta | meta-llama/llama-4-maverick, meta-llama/llama-4-scout | | Qwen | qwen/qwen3-235b-a22b, qwen/qwen3-32b | | Mistral | mistralai/mistral-large-2411 |

Pricing

You pay your LLM provider directly at their normal rates. OpenCompress charges 20% of the tokens you save — if compression saves you $1.00 in tokens, you pay us $0.20. Net saving: $0.80.

$1.00 free credit on sign-up covers ~50-100 compressed calls.

Configuration

| Key | Default | Description | |-----|---------|-------------| | apiKey | — | Your sk-occ-... key (set during onboard) | | baseUrl | https://www.opencompress.ai/api | Custom API endpoint |

Uninstall

openclaw plugins uninstall opencompress

License

MIT