kRouter
You want the hardening — verify-403 fix, MITM stream recovery, atomic backoff, SSRF guards — without paying for it.
Compare
Honest, side-by-side.
All five projects share roots in the same problem space: route LLM traffic between providers. They diverge sharply on what they prioritize. Pick by fit, not loyalty.
Cost & license
License
kRouterMIT
9routerMIT
OmniRouteMIT
LiteLLMMIT
OpenRouterClosed source
Self-hosted (no data leaves your machine)
kRouter
9router
OmniRoute
LiteLLM
OpenRouterCloud only
Cost of the tool itself
kRouterFree
9routerFree
OmniRouteFree
LiteLLMFree
OpenRouterPer-request markup
Antigravity ban handling
Bootstrap metadata enum type
kRouterNumeric (matches binary)
9routerString (triggers verify 403)
OmniRouteNumeric
LiteLLMn/a
OpenRoutern/a
Permanent-ban classifier
kRouter24h modelLock___all
9routerGeneric 1h cooldown
OmniRoute
LiteLLM
OpenRouter
Test-Connection clears every lock
kRouter
9router
OmniRoute
LiteLLM
OpenRouter
Quota tracking accuracy
Live per-model quota %
kRouter
9router`||0` paints exhausted as 100% used
OmniRoute
LiteLLM
OpenRouter
Exhausted vs 0% remaining distinction
kRouterAmber 'Exhausted • resets in X'
9routerFake red 100% bar
OmniRoute
LiteLLM
OpenRouter
TPM vs daily-quota disambiguation on 429
kRouter90s TPM cooldown
9router30m flat
OmniRoute
LiteLLM
OpenRouter
Auto-extract wrapped Antigravity usage tokens
kRouterAuto-backfills historical 0/0 rows
9routerAll rows record 0/0
OmniRoute
LiteLLM
OpenRouter
Routing & retry
Failover engine intelligence
kRouterZenith AI scoring (latency + quota mathematical ranking)
9routerDumb fill-first sequential cascade
OmniRouteSequential cascade
LiteLLMPriority sequential cascade
OpenRoutern/a
Per-provider concurrency limits
kRouterKiro 4 · Claude 5 · Antigravity 2
9routerFlat
OmniRouteTunable
LiteLLMTunable
OpenRouterProvider-side
Failover latency (RAM layer vs SQLite)
kRouter< 1ms (Sub-5ms HealthCache RAM layer)
9router~50ms disk lock per failure
OmniRoute~10ms edge state
LiteLLM~5ms Redis lock
OpenRoutern/a
Atomic backoff under concurrency
kRouterSQLite transaction
9routerRace loses increments
OmniRoute
LiteLLM
OpenRouter
MITM intercept (IDE traffic)
MITM stream-layer reliability
kRouterBounds-checked frames, NGHTTP2 → HTTP/1.1 fallback
9routerStream errors as 'Truncated event message'
OmniRouteDifferent design (edge/Deno)
LiteLLM
OpenRouter
Antigravity / Kiro / Cursor IDE intercept
kRouter
9router
OmniRoute
LiteLLM
OpenRouter
x-request-source anti-loop marker preserved
kRouter
9router
OmniRoute
LiteLLM
OpenRouter
Provider features
Thinking config preservation through blacklist
kRouterTranslates Claude/OpenAI shape
9routerBlacklist strips, never runs
OmniRoute
LiteLLM
OpenRouter
ZWJ obfuscator skips binary fields (images)
kRouter
9routerCorrupts inline_data → 400
OmniRoute
LiteLLM
OpenRouter
Number of supported providers
kRouter98+
9router40+
OmniRoute35+
LiteLLM100+
OpenRouter100+
Security
SSRF guard on baseUrl (blocks cloud metadata)
kRouter
9router
OmniRouteEdge sandbox
LiteLLM
OpenRouter
Timing-safe CLI token compare
kRouter
9router
OmniRoute
LiteLLM
OpenRouter
Per-IP brute-force lockout
kRouter
9router
OmniRoute
LiteLLM
OpenRouter
Deployment
NPM package
kRouter@sifxprime/krouter
9router9router
OmniRoute—
LiteLLMlitellm
OpenRouter—
Docker image
kRoutersifxprime/krouter
9routerdecolua/9router
OmniRoute
LiteLLMghcr.io/berriai/litellm
OpenRouter
Update cadence
kRouterHardening fixes 24-72h after report
9routerActive feature dev
OmniRouteActive independent dev
LiteLLMActive
OpenRouterActive
Which is right for you?
The honest take.
9router
You want the upstream and don't mind hitting the bugs the fork has already patched. Fastest feature cadence.
OmniRoute
You prefer an edge/Deno deployment story and a different MITM stack. Comparable hardening philosophy.
LiteLLM
You need 100+ provider breadth and don't care about IDE intercept. Enterprise SDK pattern.
OpenRouter
You want zero ops, are fine paying a per-request markup, and don't need self-hosting.