ALP AI Model Hub
Curated marketplace of ALP-optimized AI models for code review, test generation, documentation, and general-purpose task execution.
Pricing
| Plan | Price | Includes |
|---|---|---|
| Free | $0 | Browse and invoke public models |
| Pro | 15% fee on usage | Private model hosting, A/B testing, routing |
Feature Deep-Dive
Task-Optimized Models
Models fine-tuned or routed for specific ALP task types: code review, test generation, documentation, data extraction, and planning. Each model card includes latency, cost, and quality benchmarks.
A/B Testing
Route a percentage of tasks to two models and compare output quality, token cost, and latency. Statistical significance testing is built in.
Cost Optimization Routing
ALP automatically selects the cheapest model that meets quality thresholds. Fallback chains handle model outages or rate limits without human intervention.
Custom Model Registration
Register your own fine-tuned models. The hub handles versioning, A/B rollout, and metering via the @swarm_marketplace economy.
Performance Tracking
Every model invocation is recorded in the ALP timeline. Track p50/p95 latency, hallucination rates, and user feedback over time.
Use Cases
- Teams that want best-of-breed models without vendor lock-in.
- Marketplace operators listing ALP-optimized models to the ecosystem.
- Finance teams optimizing model spend via automatic routing.
Integration
Integrates with @agent model config, @swarm_marketplace metering, and @policy governance. Models are invoked as ALP tasks with full observability.
Quickstart
- Browse models:
alp hub search --task code-review - Invoke a model:
alp run --model alp/code-review-v2 --input src/ - Compare models:
alp hub ab --model-a alp/code-review-v1 --model-b alp/code-review-v2 - Register your model:
alp hub publish --model ./my-model --price 0.001/token