$ modelux vs portkey
Modelux vs Portkey
Closest peer in the space. Both route LLM traffic and provide observability. The differences are positioning and pricing: Modelux frames itself as a control plane (versioning, replay, audit, custom rule DSL) with flat-tier subscription pricing; Portkey frames itself as an AI gateway with usage-based pricing.
pick modelux when
- ▸ You want predictable flat-tier pricing as you scale
- ▸ Replay-against-history is a critical workflow for you
- ▸ You want a custom rule DSL with budget/tag awareness
- ▸ You're already using MCP-enabled coding agents
- ▸ You like the "control plane" mental model
pick portkey when
- ▸ You prefer paying per-request as you grow
- ▸ Their integration ecosystem matches your stack
- ▸ You're already using their prompt management
# compare
Capability comparison
| Capability | Modelux | Portkey |
|---|---|---|
| Positioning | ||
| Self-described as | LLM control plane | AI gateway |
| Pricing model | Flat tiers, no per-token markup | Usage-based pricing |
| Routing | ||
| Single / fallback | ✓ | ✓ |
| Cost-optimized | ✓ | ✓ |
| Latency-optimized | ✓ | ✓ |
| Ensembles + cascades | ✓ | Limited |
| Custom rule DSL | ✓ | Conditional routing |
| Control plane | ||
| Versioned configs + rollback | ✓ | Limited |
| Decision trace per request | ✓ | Logs |
| Replay simulator | ✓ | — |
| Budgets + auto-downgrade | ✓ | Caps |
| Audit log | ✓ | Limited |
| Observability | ||
| Searchable logs | ✓ | ✓ |
| Per-request cost | ✓ | ✓ |
| Latency percentiles | ✓ | ✓ |
| Warehouse export | ✓ (S3 Parquet) | ✓ |
| Pricing | ||
| Markup on tokens | None | Per-request fee on paid tiers |
| Free tier monthly volume | 10,000 requests | Limited |
| AI-native | ||
| MCP server | ✓ (80+ tools) | — |
| Public REST + OpenAPI | ✓ | ✓ |
Positioning
Self-described as
Modelux
LLM control plane
Portkey
AI gateway
Pricing model
Modelux
Flat tiers, no per-token markup
Portkey
Usage-based pricing
Routing
Single / fallback
Modelux
✓
Portkey
✓
Cost-optimized
Modelux
✓
Portkey
✓
Latency-optimized
Modelux
✓
Portkey
✓
Ensembles + cascades
Modelux
✓
Portkey
Limited
Custom rule DSL
Modelux
✓
Portkey
Conditional routing
Control plane
Versioned configs + rollback
Modelux
✓
Portkey
Limited
Decision trace per request
Modelux
✓
Portkey
Logs
Replay simulator
Modelux
✓
Portkey
—
Budgets + auto-downgrade
Modelux
✓
Portkey
Caps
Audit log
Modelux
✓
Portkey
Limited
Observability
Searchable logs
Modelux
✓
Portkey
✓
Per-request cost
Modelux
✓
Portkey
✓
Latency percentiles
Modelux
✓
Portkey
✓
Warehouse export
Modelux
✓ (S3 Parquet)
Portkey
✓
Pricing
Markup on tokens
Modelux
None
Portkey
Per-request fee on paid tiers
Free tier monthly volume
Modelux
10,000 requests
Portkey
Limited
AI-native
MCP server
Modelux
✓ (80+ tools)
Portkey
—
Public REST + OpenAPI
Modelux
✓
Portkey
✓
Based on Portkey's public documentation as of April 2026. Capability marked "Limited" means present but not as central to the product as in Modelux. If something here is wrong, email hello@modelux.ai.
Flat tiers, no per-token surprise.
Free tier covers 10k requests/month. $49/month for 100k. $199/month for 1M.
Get started free