Public betaOpen source. Built in the open.Explore the beta
OmniSwitch
QuickstartGet started
OmniSwitch / Compare

Choose for your architecture.
Not a checklist.

A practical comparison of self-hosted control planes, universal AI gateways, and governed agent infrastructure. Different operating models. Different trade-offs.

Capabilities, with the boundaries left in.

States describe the cited capability—not full product parity. Edition, deployment, and protocol scope matter. Open any state for its evidence.

Source notes
READING THE MATRIXSupportedManaged onlyPartialNot availableNot verified
Dated AI gateway capability matrix, reviewed 11 September 2026
CapabilityOmniSwitch[1]Register a8a594bLiteLLM[2]Python proxy; edition scope variesPortkey[3]OSS gateway; cloud shown separatelyHelicone[4]OSS gateway; maintenance contextKong AI Gateway[5]OSS ai-proxy scope; AI license variesCloudflare AI Gateway[6]Managed serviceAgent Router[7]Formerly Envoy AI Gateway
Self-hosted deployment
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Routing & fallbacks
Supported

route-level fallbacks, max_retries, retry_backoff, retry_codes, per-attempt timeout.

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗
Supported

Documented fallback and retry configuration across deployments.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11
Open cited source ↗
Supported

Fallback, retries, and timeouts documented.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11
Open cited source ↗
Supported

Failover documented as a core feature.

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11
Open cited source ↗
Not available

Multi-provider fallback across models/providers is an AI Proxy Advanced (Enterprise) capability, not the OSS AI Proxy plugin used here.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11
Open cited source ↗
Managed only

Request retries and model fallback documented.

Version
hosted service, n/a
Verified
2026-09-11
Open cited source ↗
Supported

Provider fallback / automatic failover documented.

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11
Open cited source ↗
MCP gateway
Supported

HTTP MCP targets with streamed SSE / streamable-HTTP pass-through.

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗
Supported

MCP gateway support documented.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11
Open cited source ↗
Supported

MCP Gateway for remote MCP server connectivity documented.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11
Open cited source ↗
Not verified

No evidence of MCP support found in available sources; unconfirmed, not necessarily absent.

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Supported

Kong documents an MCP Gateway / MCP server breakdown as part of the AI Gateway line.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11
Open cited source ↗
Not verified

Managed service: capability-map only, no same-host benchmark run. Not verified in available sources.

Version
hosted service, n/a
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Supported

Full MCP gateway with server multiplexing, tool routing/filtering, OAuth, and observability documented.

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11
Open cited source ↗
A2A task methods
Partial

SendMessage / SendMessageStream, GetTask, CancelTask implemented (JSON-RPC methods and REST aliases). Agent card advertises pushNotifications: false, so push-notification task updates are not supported.

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗

Current protocol scope: SendMessage, SendMessageStream (SSE task events), GetTask, best-effort CancelTask; tasks in memory; no push notifications; no downstream agent proxy.

SendMessage / SendMessageStream, GetTask, CancelTask implemented (JSON-RPC methods and REST aliases). Agent card advertises pushNotifications: false, so push-notification task updates are not supported.

Not available

No A2A task lifecycle found in available sources.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not available

This project's internal comparison table lists Portkey as "No" for A2A task lifecycle.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11
Open cited source ↗
Not available

No A2A support found in available sources.

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not available

This project's internal comparison table lists Kong as "No" for A2A task lifecycle.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11
Open cited source ↗
Not verified

Managed service: capability-map only, no same-host benchmark run. Not verified in available sources.

Version
hosted service, n/a
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not available

Same as above.

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Token-based rate limits
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Built-in guardrails
Supported

Built-in PII/prompt-injection/SQL-injection/toxicity/secret detectors plus custom regex rules, single-pass Rust-WASM scan with pure-Go fallback.

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗
Supported

Guardrail hooks including PII masking (Presidio) and moderation.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11
Open cited source ↗
Supported

Integrated guardrails documented.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11
Open cited source ↗
Not verified

Security/request validation documented; no formal "guardrails" product label found in the strongest available sources.

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Partial

AI Prompt Guard (allow/deny lists) is documented as available; AI Semantic Prompt Guard and vendor guardrail integrations (Bedrock/Azure/Model Armor) are Enterprise/paid per third-party summaries, not independently re-verified per-plugin.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11
Open cited source ↗

AI Prompt Guard (allow/deny lists) is documented as available; AI Semantic Prompt Guard and vendor guardrail integrations (Bedrock/Azure/Model Armor) are Enterprise/paid per third-party summaries, not independently re-verified per-plugin.

Managed only

Guardrails evaluate prompts/responses using @cf/meta/llama-guard-3-8b on Workers AI.

Version
hosted service, n/a
Verified
2026-09-11
Open cited source ↗
Not verified

Not confirmed in available sources.

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
OTLP telemetry export
Supported

OTel trace export (tested against Langfuse/Jaeger/Tempo).

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗
Supported

OpenTelemetry callback integration documented.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11
Open cited source ↗
Not verified

OTel ingestion is documented for the Portkey platform; OSS-gateway-only export not independently confirmed.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

Not confirmed in available sources.

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

Kong Gateway has general OTel plugin support; AI-traffic-specific OTLP export was not independently confirmed.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Not verified

Analytics/logging documented; OTLP export specifically not confirmed.

Version
hosted service, n/a
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Supported

Enterprise observability documented; Envoy's ecosystem is commonly OTel-instrumented (e.g. Apache SkyWalking's Envoy AI Gateway monitoring guide).

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11
Open cited source ↗
License
Supported

Apache-2.0.

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗
Partial

MIT for the core; the enterprise/ directory is separately licensed (proprietary) per the repo's own LICENSE file.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11
Open cited source ↗

MIT for the core; the enterprise/ directory is separately licensed (proprietary) per the repo's own LICENSE file.

Supported

MIT.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11
Open cited source ↗
Supported

GPL-3.0 (verified via the repository's GitHub license metadata; some third-party articles incorrectly state Apache-2.0/MIT).

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11
Open cited source ↗
Supported

Apache-2.0 for Kong Gateway OSS; several AI Gateway plugins used for full parity (AI Proxy Advanced, AI Semantic Cache tier, AI Rate Limiting Advanced, AI Semantic Prompt Guard) are Enterprise-licensed and are marked 'Enterprise, not tested' in this harness rather than benchmarked.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11
Open cited source ↗
Not available

Closed managed service; not self-hostable/open-source.

Version
hosted service, n/a
Verified
2026-09-11
Open cited source ↗
Not verified

The former envoyproxy/ai-gateway GitHub repository now redirects to theagentrouter/agent-router (project renamed to "Agent Router" during 2026; the project states images, CRDs, CLI (aigw), and Go module path are unchanged). License metadata read back from the redirect target (Apache-2.0) but this may reflect a different/renamed project's repo settings rather than a re-verified license for the exact codebase; treat as unknown pending direct confirmation.

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11
Open cited source ↗

Self-hosted deployment

OmniSwitch[1]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
LiteLLM[2]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Portkey[3]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Helicone[4]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Kong AI Gateway[5]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Cloudflare AI Gateway[6]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Agent Router[7]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗

Routing & fallbacks

OmniSwitch[1]
Supported

route-level fallbacks, max_retries, retry_backoff, retry_codes, per-attempt timeout.

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗
LiteLLM[2]
Supported

Documented fallback and retry configuration across deployments.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11
Open cited source ↗
Portkey[3]
Supported

Fallback, retries, and timeouts documented.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11
Open cited source ↗
Helicone[4]
Supported

Failover documented as a core feature.

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11
Open cited source ↗
Kong AI Gateway[5]
Not available

Multi-provider fallback across models/providers is an AI Proxy Advanced (Enterprise) capability, not the OSS AI Proxy plugin used here.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11
Open cited source ↗
Cloudflare AI Gateway[6]
Managed only

Request retries and model fallback documented.

Version
hosted service, n/a
Verified
2026-09-11
Open cited source ↗
Agent Router[7]
Supported

Provider fallback / automatic failover documented.

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11
Open cited source ↗

MCP gateway

OmniSwitch[1]
Supported

HTTP MCP targets with streamed SSE / streamable-HTTP pass-through.

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗
LiteLLM[2]
Supported

MCP gateway support documented.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11
Open cited source ↗
Portkey[3]
Supported

MCP Gateway for remote MCP server connectivity documented.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11
Open cited source ↗
Helicone[4]
Not verified

No evidence of MCP support found in available sources; unconfirmed, not necessarily absent.

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Kong AI Gateway[5]
Supported

Kong documents an MCP Gateway / MCP server breakdown as part of the AI Gateway line.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11
Open cited source ↗
Cloudflare AI Gateway[6]
Not verified

Managed service: capability-map only, no same-host benchmark run. Not verified in available sources.

Version
hosted service, n/a
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Agent Router[7]
Supported

Full MCP gateway with server multiplexing, tool routing/filtering, OAuth, and observability documented.

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11
Open cited source ↗

A2A task methods

OmniSwitch[1]
Partial

SendMessage / SendMessageStream, GetTask, CancelTask implemented (JSON-RPC methods and REST aliases). Agent card advertises pushNotifications: false, so push-notification task updates are not supported.

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗

Current protocol scope: SendMessage, SendMessageStream (SSE task events), GetTask, best-effort CancelTask; tasks in memory; no push notifications; no downstream agent proxy.

SendMessage / SendMessageStream, GetTask, CancelTask implemented (JSON-RPC methods and REST aliases). Agent card advertises pushNotifications: false, so push-notification task updates are not supported.

LiteLLM[2]
Not available

No A2A task lifecycle found in available sources.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Portkey[3]
Not available

This project's internal comparison table lists Portkey as "No" for A2A task lifecycle.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11
Open cited source ↗
Helicone[4]
Not available

No A2A support found in available sources.

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Kong AI Gateway[5]
Not available

This project's internal comparison table lists Kong as "No" for A2A task lifecycle.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11
Open cited source ↗
Cloudflare AI Gateway[6]
Not verified

Managed service: capability-map only, no same-host benchmark run. Not verified in available sources.

Version
hosted service, n/a
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Agent Router[7]
Not available

Same as above.

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗

Token-based rate limits

OmniSwitch[1]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
LiteLLM[2]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Portkey[3]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Helicone[4]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Kong AI Gateway[5]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Cloudflare AI Gateway[6]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Agent Router[7]
Not verified

No claim is recorded for this capability in the pinned register.

Version
Not recorded
Verified
Not recorded

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗

Built-in guardrails

OmniSwitch[1]
Supported

Built-in PII/prompt-injection/SQL-injection/toxicity/secret detectors plus custom regex rules, single-pass Rust-WASM scan with pure-Go fallback.

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗
LiteLLM[2]
Supported

Guardrail hooks including PII masking (Presidio) and moderation.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11
Open cited source ↗
Portkey[3]
Supported

Integrated guardrails documented.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11
Open cited source ↗
Helicone[4]
Not verified

Security/request validation documented; no formal "guardrails" product label found in the strongest available sources.

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Kong AI Gateway[5]
Partial

AI Prompt Guard (allow/deny lists) is documented as available; AI Semantic Prompt Guard and vendor guardrail integrations (Bedrock/Azure/Model Armor) are Enterprise/paid per third-party summaries, not independently re-verified per-plugin.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11
Open cited source ↗

AI Prompt Guard (allow/deny lists) is documented as available; AI Semantic Prompt Guard and vendor guardrail integrations (Bedrock/Azure/Model Armor) are Enterprise/paid per third-party summaries, not independently re-verified per-plugin.

Cloudflare AI Gateway[6]
Managed only

Guardrails evaluate prompts/responses using @cf/meta/llama-guard-3-8b on Workers AI.

Version
hosted service, n/a
Verified
2026-09-11
Open cited source ↗
Agent Router[7]
Not verified

Not confirmed in available sources.

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗

OTLP telemetry export

OmniSwitch[1]
Supported

OTel trace export (tested against Langfuse/Jaeger/Tempo).

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗
LiteLLM[2]
Supported

OpenTelemetry callback integration documented.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11
Open cited source ↗
Portkey[3]
Not verified

OTel ingestion is documented for the Portkey platform; OSS-gateway-only export not independently confirmed.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Helicone[4]
Not verified

Not confirmed in available sources.

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Kong AI Gateway[5]
Not verified

Kong Gateway has general OTel plugin support; AI-traffic-specific OTLP export was not independently confirmed.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Cloudflare AI Gateway[6]
Not verified

Analytics/logging documented; OTLP export specifically not confirmed.

Version
hosted service, n/a
Verified
2026-09-11

No capability-specific source was recorded. The register is the provenance, not proof of absence.

Open pinned register ↗
Agent Router[7]
Supported

Enterprise observability documented; Envoy's ecosystem is commonly OTel-instrumented (e.g. Apache SkyWalking's Envoy AI Gateway monitoring guide).

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11
Open cited source ↗

License

OmniSwitch[1]
Supported

Apache-2.0.

Version
v0.2.0-beta (commit 88a27c0)
Verified
2026-09-11
Open cited source ↗
LiteLLM[2]
Partial

MIT for the core; the enterprise/ directory is separately licensed (proprietary) per the repo's own LICENSE file.

Version
ghcr.io/berriai/litellm:main-stable @ sha256:a3715fa7ad8387941ab697259bd2881d68931657247a41984f90fae6d11c62bf
Verified
2026-09-11
Open cited source ↗

MIT for the core; the enterprise/ directory is separately licensed (proprietary) per the repo's own LICENSE file.

Portkey[3]
Supported

MIT.

Version
latest @ docker.io/portkeyai/gateway:latest sha256:97f094d9c8a764cbfaa2a7138c0017b247ca923bb06db1b4c13b7f8a33b5200d
Verified
2026-09-11
Open cited source ↗
Helicone[4]
Supported

GPL-3.0 (verified via the repository's GitHub license metadata; some third-party articles incorrectly state Apache-2.0/MIT).

Version
latest @ docker.io/helicone/ai-gateway:latest sha256:bdf1f7c4f3aa177b8a03a02d461c744733c0292afd95d3185337d9fc2c346af6
Verified
2026-09-11
Open cited source ↗
Kong AI Gateway[5]
Supported

Apache-2.0 for Kong Gateway OSS; several AI Gateway plugins used for full parity (AI Proxy Advanced, AI Semantic Cache tier, AI Rate Limiting Advanced, AI Semantic Prompt Guard) are Enterprise-licensed and are marked 'Enterprise, not tested' in this harness rather than benchmarked.

Version
Kong Gateway 3.9 (OSS) + ai-proxy plugin @ docker.io/library/kong:3.9 sha256:ca71c5591eabaf18de96d26b7eed5e2fdb590dac141e467a779c38017e5bdf81
Verified
2026-09-11
Open cited source ↗
Cloudflare AI Gateway[6]
Not available

Closed managed service; not self-hostable/open-source.

Version
hosted service, n/a
Verified
2026-09-11
Open cited source ↗
Agent Router[7]
Not verified

The former envoyproxy/ai-gateway GitHub repository now redirects to theagentrouter/agent-router (project renamed to "Agent Router" during 2026; the project states images, CRDs, CLI (aigw), and Go module path are unchanged). License metadata read back from the redirect target (Apache-2.0) but this may reflect a different/renamed project's repo settings rather than a re-verified license for the exact codebase; treat as unknown pending direct confirmation.

Version
Agent Router (formerly Envoy AI Gateway); aigw CLI image docker.io/envoyproxy/ai-gateway-cli:latest sha256:ad0d4ffe2924dd3f46414687e9331f5fa947b9f2c9db735accaad9f2275aff56
Verified
2026-09-11
Open cited source ↗

OmniSwitch scope. The pinned register marks OpenAI Responses API, Anthropic Messages API, Native (non-passthrough) provider adapters, Rate limiting, Mutual TLS, OpenAPI-to-MCP conversion, A2A task methods, Prompt management, Evals / experiments, Published cross-gateway benchmarks as partial or unavailable. Read each source and version; compare them with the build you actually deploy.

Read unknowns carefully. “Not verified” means an evidence gap, not absence. Editions remain separate. Register a8a594b; 7 compared deployments.

A2A (Beta). SendMessage, SendMessageStream (SSE task events), GetTask, best-effort CancelTask; tasks in memory; no push notifications; no downstream agent proxy.

When to choose which

Start with the constraint you can’t move.

Operating model, ecosystem, governance, or an existing investment—not a universal winner.

Where we fit

Choose OmniSwitch for
self-hosted control.

A compact control layerA single Go binary, local policy checks, scoped keys, budgets, and request traces. A fit when you want to inspect and operate the gateway yourself.

Know the current limitsPublic beta, single-replica SQLite, and documented compatibility API subsets. No dashboard SSO or admin audit trail yet. Validate unreleased work separately.

LiteLLM[2]

Provider breadth and a Python-first ecosystem.

Evaluate its provider integrations and operational ecosystem. Review the documented Enterprise boundary for admin audit logs and SSO beyond five users.

Read the full comparison

Portkey[3]

An OSS gateway with a wider platform path.

Evaluate its routing and platform workflow. Check the current Enterprise boundary for SSO, SCIM, and audit logs rather than relying on older SaaS-only comparisons.

Read the full comparison

Helicone[4]

Your existing observability workflow still fits.

Staying can be reasonable: services remain live in maintenance mode. Confirm roadmap and support expectations before committing to a new dependency.

Read the full comparison

Kong AI Gateway[5]

AI governance inside your API platform.

Evaluate Kong when its API platform is already part of your architecture. Advanced AI plugins have licensing boundaries; distinguish them from OSS ai-proxy capabilities.

Read the full comparison

Cloudflare AI Gateway[6]

A managed gateway at your existing edge.

Evaluate it when you prefer a managed service and already use Cloudflare. Compare the network and control-plane model separately from a local self-hosted binary.

Read the full comparison

Agent Router[7]

Kubernetes and Envoy are already your foundation.

Evaluate its Kubernetes/Envoy integration when that platform matches your operating model. Agent Router is the renamed Envoy AI Gateway, not the separate AgentGateway project.

Read the full comparison
Migration starting points

Move one workflow. Verify the behavior.

A base URL change is a start—not a complete migration plan.

From LiteLLM

Pilot the client contract before changing traffic.

  1. Inventory routes, key scopes, callbacks, and metadata.
  2. Map routing, fallback, and budget settings explicitly.
  3. Verify representative chat requests, streaming, tools, and errors.
Pilot one LiteLLM workflow
From Portkey

Translate the configuration. Don’t assume it transfers.

  1. Inventory configs, gateway keys, provider credentials, and policies.
  2. Map retries, routing, budgets, and guardrails.
  3. Verify request shaping, streaming, and error handling against your app.
Pilot one Portkey workflow
From Helicone

Keep the observability questions. Re-map the gateway contract.

  1. Inventory routes, custom headers, and metadata.
  2. Map keys, fallback behavior, and log fields.
  3. Pilot representative calls and inspect traces before moving traffic.
Pilot one Helicone workflow

No competitor-header compatibility or automatic configuration/key import is implied. Keep your current gateway available until the pilot meets your requirements.

The market moves. Context matters.

Ownership and project names change; evaluate the current product and support model.

As of 9 October 2026

Helicone → Mintlify

Helicone announced it was joining Mintlify on 3 March 2026. Services remain live in maintenance mode, with security updates, new models, and fixes continuing.

Source · accessed 2026-10-09

Portkey → Palo Alto Networks

Palo Alto Networks completed its acquisition of Portkey on 29 May 2026. Review the current product editions and terms, not older SaaS-only comparisons.

Source · accessed 2026-10-09

Envoy AI Gateway → Agent Router

The project was renamed Agent Router in v1.2.0 on 6 October 2026. Existing API groups, CRDs, charts, images, and CLI names remain unchanged. It is not the separate AgentGateway project.

Source · accessed 2026-10-09

Sources & methodology

Register verification dates are shown per cell. Editorial context accessed 9 October 2026. Not independently tested parity.

Footnotes [1]–[7]
  1. [1]

    OmniSwitch

    Product register a8a594b. Cells reflect registered source claims, not roadmap plans or release-artifact availability; each retains its own verified date and version.

    Read 5 cited sources
  2. [2]

    LiteLLM

    Python proxy; edition scope varies. Capability-specific documentation and dated editorial sources remain separate from independent testing.

    Read 11 cited sources
  3. [3]

    Portkey

    OSS gateway; cloud shown separately. Capability-specific documentation and dated editorial sources remain separate from independent testing.

    Read 8 cited sources
  4. [4]

    Helicone

    OSS gateway; maintenance context. Capability-specific documentation and dated editorial sources remain separate from independent testing.

    Read 5 cited sources
  5. [5]

    Kong AI Gateway

    OSS ai-proxy scope; AI license varies. Capability-specific documentation and dated editorial sources remain separate from independent testing.

    Read 11 cited sources
  6. [6]

    Cloudflare AI Gateway

    Managed service. Capability-specific documentation and dated editorial sources remain separate from independent testing.

    Read 5 cited sources
  7. [7]

    Agent Router

    Formerly Envoy AI Gateway. Capability-specific documentation and dated editorial sources remain separate from independent testing.

    Read 6 cited sources

Supported means documented within the named capability and deployment scope, not full product parity. Beta and licensed editions are shown when the register supplies a qualifier. Not verified describes an evidence gap, not absence. Claims come from the pinned register; cells have no scores or totals. This is a dated documentation review, not a performance test or a security certification. See the benchmark methodology.

Does self-hosted control fit your stack?

Start with a local gateway and one representative request.

Run OmniSwitchRead the docs