Skip to content

Scan Criteria

What constitutes a competitive signal for OpenRouter. Referenced by the scan and audit-sources skills. Update this file as our competitive priorities evolve.

Last reviewed: 2026-03-26

Competitive Signals (report these)

  • Model/provider additions or removals (new models, deprecated models, new provider integrations)
  • Pricing changes (new tiers, price adjustments, credit/bonus changes, markup changes, BYOK pricing)
  • API capabilities (new endpoints, new protocols like gRPC/WebSocket, new SDK features, breaking changes)
  • Gateway features: caching, rate limiting, load balancing, fallbacks, retries, request routing
  • Observability features: logging, tracing, analytics, dashboards, cost tracking
  • Enterprise features: SSO, RBAC, compliance certifications, audit logs, SLAs, private deployments
  • Agent/MCP features: MCP server support, agent frameworks, tool management, function calling
  • SDK releases that affect gateway adoption (new language support, major versions, breaking changes)
  • Competitive positioning moves: unified billing, marketplace launches, acquisition announcements
  • Infrastructure changes relevant to AI workloads: new regions, edge deployments, compute tiers

Noise (skip these — update snapshots but no changelog entry)

  • Generic cloud/platform updates unrelated to AI gateway (CDN, DNS, generic storage)
  • Marketing copy changes, testimonials, hero images, customer logo updates
  • Template/starter project releases (unless revealing new gateway capabilities)
  • UI component library updates (unless they're gateway UI features)
  • Minor doc formatting, typo fixes, navigation restructuring
  • Blog posts about general AI trends that don't announce product features or capabilities
  • Conference sponsorships, hiring announcements, community events
  • Changes to products outside the competitor's AI gateway/routing line

Important: Blog posts that announce new product features, APIs, pricing changes, or capabilities are signals, not noise — even when found on a company-wide blog. The "general AI trends" exclusion above applies only to thought-leadership / opinion pieces with no product announcements. When in doubt, treat a blog post as signal.

High-Value Source Characteristics

A good monitoring source: - Documents specific product capabilities (API reference, feature docs, configuration options) - Lists supported models, providers, or integrations (changes here = direct competitive signal) - Shows pricing (changes here = high-impact) - Has a product changelog (concentrated signal, low noise) - Returns structured/machine-readable data (JSON API endpoints for model lists) - Returns meaningful content without JavaScript rendering

A poor monitoring source: - Generic landing page with marketing copy (changes = noise) - Top-level docs index with only links (no substantive content to diff) - Pages that require JS to render (empty/skeleton responses)

Note on company blogs: Company-wide blogs have a low signal-to-noise ratio, but they are often the first place major product announcements appear (new APIs, pricing changes, feature launches). Do NOT dismiss blog changes as noise without checking whether they announce product capabilities listed in "Competitive Signals" above. A blog post announcing a new API or feature is high-value signal regardless of the source's overall noise level.