neurometric / ClawPack

SLM bundles built for OpenClaw

ClawPack gives your agentic workflows a team of task-specific specialists instead of one expensive generalist. Drop-in compatible. Download free or host for $2/mo per model, unlimited.

Coming soon
6
Pre-built bundles
34
Task-specific SLMs
10–50×
Cheaper per sub-task
<60ms
Median latency
How It Works

Swap frontier model calls for specialists

In a typical OpenClaw agent graph, every sub-task routes to the same frontier model. ClawPack replaces each node with a trained specialist.

1

Pick a bundle

Choose a pre-built ClawPack — Legal, Finance, Sales, Support, Coding, or Marketing — or mix models from different bundles.

2

Wire into OpenClaw

Each SLM registers as a tool node in OpenClaw's agent graph. One config file. No custom code required. Runs alongside your existing tools.

3

Run locally or hosted

Download models and run them on your own hardware for free, or deploy hosted and get unlimited inference at $2/mo per model.

Bundles

Six packs. Thirty-four specialists.

Each ClawPack bundle is a curated set of SLMs that cover an end-to-end workflow. Use one bundle or mix models across several.

5 models
Legal Review
End-to-end contract analysis — risk extraction, clause parsing, redline diffs, compliance checks, memo drafting.
Included Models
contract-risk-analyzer3.2B
clause-extraction-engine2.1B
redline-diff-summarizer1.8B
compliance-flag-checker2.4B
legal-memo-drafter7B
$10 /mo for all 5
6 models
Financial Analysis
Invoice extraction, expense categorization, financial statement parsing, ratio analysis, forecast narratives.
Included Models
invoice-data-extractor2.8B
expense-categorizer1.5B
financial-stmt-parser3.8B
ratio-analyzer2.2B
forecast-narrator7B
anomaly-flagger1.8B
$12 /mo for all 6
6 models
Coding Swarm
Write, review, test, document, scan, and refactor — a coordinated team for every step of the dev lifecycle.
Included Models
code-writer7B
code-reviewer3B
test-generator3B
doc-writer3B
security-scanner2.4B
code-refactorer3B
$12 /mo for all 6
6 models
Sales Enablement
Cold outreach, lead scoring, objection handling, proposal drafting, meeting prep, and CRM summarization.
Included Models
cold-outreach-writer7B
lead-scorer1.5B
objection-handler3B
proposal-drafter7B
meeting-prep-agent3B
crm-summarizer2.1B
$12 /mo for all 6
5 models
Customer Ops
Ticket triage, sentiment detection, response drafting, escalation routing, and knowledge base search.
Included Models
ticket-triage-router1.5B
sentiment-detector1.2B
response-drafter7B
escalation-classifier1.5B
kb-retrieval-ranker2.4B
$10 /mo for all 5
6 models
Marketing & Content
Ad copy, SEO briefs, social posts, email sequences, headline testing, and brand voice matching.
Included Models
ad-copy-generator7B
seo-brief-writer3B
social-post-creator3B
email-sequence-writer7B
headline-tester1.5B
brand-voice-matcher2.4B
$12 /mo for all 6
Coming soon
Integration

Wires into OpenClaw in five minutes

Each ClawPack model registers as a standard tool node in OpenClaw's agent graph. Your orchestrator calls it like any other tool — same interface, dramatically lower cost.

  • ✓OpenAI-compatible endpoint — works with any framework that speaks /v1/chat/completions
  • ✓Runs as a local sidecar or hosted endpoint — toggle with one env variable
  • ✓Also works with LangChain, CrewAI, AutoGen, and bare HTTP calls
  • ✓Structured JSON output for every model — no parsing gymnastics
openclaw_config.yaml
# ClawPack — Legal Review Bundle
tools:
  - name: contract-risk-analyzer
    provider: neurometric
    model: clawpack/legal/risk-analyzer
    mode: hosted # or "local"

  - name: clause-extractor
    provider: neurometric
    model: clawpack/legal/clause-engine
    mode: hosted

  - name: compliance-checker
    provider: neurometric
    model: clawpack/legal/compliance
    mode: local

# That's it. Your agent graph now
# routes legal tasks to specialists.

ClawPack vs. frontier model routing

What changes when every sub-task gets its own specialist

ClawPackGPT-4o for all
Cost per 1,000 sub-tasks~$0.30$15.00
Median latency< 60ms400-800ms
Hosted pricing$2/mo unlimitedPer-token
Task-specific accuracy95%+ on taskVaries widely
Self-host optionFree downloadNot available
Data leaves your stackNever (if self-hosted)Always

Simple pricing, no token math

Download every ClawPack model for free. Host with us for flat-rate unlimited inference.

Self-hosted
Free
Download any ClawPack bundle. Run on your own hardware forever. No API key required. No telemetry.
Hosted
$2 / model / mo
Unlimited hosted inference per model. No token caps. The full Legal bundle (5 models) runs for $10/mo total. All six bundles for under $70/mo.
Enterprise
Custom
Dedicated endpoints, VPC deployment, custom SLM training, volume discounts, and forward-deployed engineering support.
Coming soon

Your agents deserve better tools

Download ClawPack right now. Wire it into OpenClaw in five minutes. Start saving immediately.

Coming soon