VerdictPal
Join
← All cardsDOSSIER · Model infrastructure · SOLID · VERIFIED 2026-06-09
OpenRouter86SolidBenchmark pendingLLM model router and cost-comparison gatewayVerified 2026-06-09

Dossier · Model infrastructure

OpenRouter

LLM model router and cost-comparison gateway · last verified 2026-06-09

Solid
Model infrastructure
OpenRouter
86/100
ROLELLM model router and cost-comparison gateway
Editorial fit
86
Source quality
72
Citation honesty
70
Privacy controls
58
Value for money
92
Speed
80
FREEFree
CUSTOMPay-as-you-go
CUSTOMEnterprise
SolidVerified 2026-06-09VP·METHOD

VERDICT HISTORY

  • Added LongCat-2.0/Owl Alpha release context: Meituan identified the OpenRouter stealth route as LongCat-backed after it had already become a high-volume coding model. Treat it as evidence of routing-market adoption, not as a privacy shortcut or independently verified benchmark row.
  • OpenRouter is the model-router card. It is useful when a developer wants one OpenAI-compatible API surface for many LLMs, fallback routing, model comparisons, BYOK, and spend controls. The updated card sharpens the key caveat: routing convenience adds a platform fee and a subprocessor/provider-retention surface that must be controlled before confidential research or user data flows through it.
  • Converted imported Notion research into a full flagship-ready dossier template with metrics, panes, pricing deck, scenarios, benchmark rows, and comparison slices.

EDITOR'S NOTE

Desk hands-on 2026-06: Incredible model switchboard — one API, sane routing, worth the platform fee for multi-model workflows.

AT A GLANCE

OpenAI-compatible model-routing API for comparing and shipping across many LLM providers, with pay-as-you-go credits, fallback routing, BYOK, activity logs, spend budgets, and enterprise data-policy controls.

Role: LLM model router and cost-comparison gatewayCategory: Model infrastructureEditorial · hands-on

OpenRouter flagship-ready dossier: LLM model router and cost-comparison gateway.

PUBLIC FACTS · vendor & repo

List prices and pay-as-you-go entry points we can cite without running our own bench. Each tile links to a source when possible.

Model infrastructureSetVerdictPal card ↗
synthesizedEvidenceQuality gate ↗
4Sources checkedCard sources ↗
2026-06-07Pricing checkedQuality gate ↗

Limits & product surface

Non-price vendor claims — multipliers, caps, and API scope. Detailed matrices live in subscription and SDK sections below.

Primary surface
apiCard identity ↗
Modalities
text, code, dataCard identity ↗
Workflow roles
[object Object], [object Object], [object Object]VerdictPal editorial ↗
Alternatives tracked
LiteLLM, Together AI, Fireworks AI, Amazon Bedrock, Direct OpenAI APIVerdictPal comparison set ↗

HOW IT WORKS · agent loop

The public positioning for this product — the loop we score against on VerdictPal.

Start with the wedge

Developers comparing model behavior and cost without rewriting API clients

Run the representative task

Run one VerdictPal summary prompt across six routed models.

Check the failure modes

Free tier capped to free models

Compare before recommending

Compare against LiteLLM, Together AI, Fireworks AI before shipping advice.

METRIC LAB · 16 DIMENSIONS

Click a tile for the editorial note. Color follows score: coral, yellow, mint.

Shape

Avg 81 · 58–98

86Editorial fit

Weighted roll-up across 13 dimensions for Model infrastructure; pending desk verification if rescored from public facts.

By group

Fit87
  • Editorial fit86
  • Wedge task fit84
  • Feature depth90
Cost84
  • Free-tier utility82
  • Cost-to-value92
  • Opportunity cost78
Trust74
  • Source grounding70
  • Privacy posture58
  • Failure transparency74
  • Evidence strength72
  • Transparency98
Workflow83
  • Integration reach94
  • Setup friction80
  • Reliability84
  • Competitive position90
  • Data portability65

SEARCH MODES · 4 lenses

One input box, many retrieval postures. Filter by tier.

01Free

REST default

REST default path for OpenRouter — verify limits on the live product.

02Pro

Batch jobs

Batch jobs path for OpenRouter — verify limits on the live product.

03Pro

Streaming

Streaming path for OpenRouter — verify limits on the live product.

04Max

Webhooks

Webhooks path for OpenRouter — verify limits on the live product.

BENCHMARK LEDGER

Public rows are vendor or third-party claims we logged with a date. Desk rows are reserved for VerdictPal self-run results.

MeasureResultSource
Run one VerdictPal summary prompt across six routed modelsoutput quality, cost, latency, provider, model ID stability, failure behavior.PlannedVerdictPal benchmark plan
Force primary provider failure and observe fallback routeroute transparency, cost change, latency, output change, retention/provider change.PlannedVerdictPal benchmark plan
Configure provider-retention preferences and send test requestsaccount-level clarity, per-call controls, logs, Enterprise data-policy routing, audit trail.PlannedVerdictPal benchmark plan
Compare OpenRouter PAYG, OpenRouter BYOK, and direct provider callscost, setup time, observability, provider control, subprocessor complexity.PlannedVerdictPal benchmark plan

PRICING DECK · checked 2026-06-09

Free

$0

free

  • 50 requests/day on free models
  • Verify live regional checkout before publishing procurement advice.
Desk pick

Pay-as-you-go

Custom

usage

  • Model list price + 5.5% platform fee
  • Verify live regional checkout before publishing procurement advice.

RAW · checked 2026-06-09: Free: 25+ free models, 4 free providers, 50 req/day · Pay-as-you-go: 400+ models, 60+ providers, 5.5% platform fee, no minimum spend · Enterprise: bulk discounts, SLA/shared Slack, optional dedicated limits · BYOK existing desk data: 1M free reqs/mo then 5% fee on paid accounts.

RESEARCH LOG · desk notes

What we learned while building this dossier — not vendor copy.

Notion Card pipeline ↗

Notion research pass

OpenRouter is the model-router card. It is useful when a developer wants one OpenAI-compatible API surface for many LLMs, fallback routing, model comparisons, BYOK, and spend controls. The updated card sharpens the key caveat: routing convenience adds a platform fee and a subprocessor/provider-retention surface that must be controlled before confidential research or user data flows through it.

VerdictPal git ↗

Flagship-ready structure

Converted imported research into metrics, panes, scenarios, benchmark rows, comparison notes, and pricing deck.

COMPETITIVE LENS

Where OpenRouter wins for cited research — and where a rival still belongs in the stack.

OpenRouter is stronger when developers comparing model behavior and cost without rewriting api clients; LiteLLM may still win for narrower fit, procurement, or specialist depth.

OpenRouter

  • Developers comparing model behavior and cost without rewriting API clients
  • Apps that need fallback routing across multiple LLM providers

LiteLLM

  • You need a single-vendor DPA with no subprocessors or upstream providers
  • You cannot monitor per-model token spend, routing, provider retention, and model deprecations

UNDER THE HOOD

Vendors named on the product about page — useful for procurement and privacy reviews.

model routing
OpenRouter
API comparison
LiteLLM
cost-sensitive model testing
Together AI

DEEP PANES · 6 LENSES

Editorial lenses only. Subscription and API pricing live in their own sections above.

Lens 01

Positioning

OpenRouter is framed here as LLM model router and cost-comparison gateway. OpenRouter is the model-router card. It is useful when a developer wants one OpenAI-compatible API surface for many LLMs, fallback routing, model comparisons, BYOK, and spend controls. The updated card sharpens the key caveat: routing convenience adds a platform fee and a subprocessor/provider-retention surface that must be controlled before confidential research or user data flows through it.

Checked
2026-06-07
Evidence
synthesized
  • Developers comparing model behavior and cost without rewriting API clients
  • Apps that need fallback routing across multiple LLM providers
  • Budget-aware experiments where per-model logs and spend controls matter
  • Teams that need one gateway while still choosing provider/model/data-policy routes deliberately

TEST SCENARIOS · hands-on lab

How we exercised the product. Step through each run before you trust the scores.

Test run

Step 1 of 4

Start from the persona in best-for item 1.

BEST FOR

  • Developers comparing model behavior and cost without rewriting API clients
  • Apps that need fallback routing across multiple LLM providers
  • Budget-aware experiments where per-model logs and spend controls matter
  • Teams that need one gateway while still choosing provider/model/data-policy routes deliberately

AVOID IF

  • You need a single-vendor DPA with no subprocessors or upstream providers
  • You cannot monitor per-model token spend, routing, provider retention, and model deprecations
  • You want a consumer chat UI rather than an API gateway
  • Your application cannot tolerate upstream provider outages, deprecations, or policy differences

STRENGTHS

  • OpenAI-compatible API
  • free model tier
  • 400+ paid models
  • 60+ providers
  • auto-routing
  • preferred vendors
  • provider data-policy tags
  • activity/input/output logs
  • spend budgets
  • BYOK with 1M free requests/month then 5% fee
  • Enterprise data-policy routing
  • model fallback
  • ZDR/provider-logging/privacy docs
  • cost/model comparison

WEAKNESSES

  • Free tier capped to free models
  • PAYG adds 5.5% platform fee
  • BYOK fee after 1M free monthly requests
  • Provider outages/deprecations/model ID changes can break integrations
  • Provider data-policy tags are not definitive legal policy sources
  • Fallback routing can silently change cost/output/provider/privacy if unconstrained

EDITORIAL EVIDENCE · 4 ENTRIES

TASK

Run one VerdictPal summary prompt across six routed models.

Pending.

↳ output quality, cost, latency, provider, model ID stability, failure behavior.

TASK

Force primary provider failure and observe fallback route.

Pending.

↳ route transparency, cost change, latency, output change, retention/provider change.

TASK

Configure provider-retention preferences and send test requests.

Pending.

↳ account-level clarity, per-call controls, logs, Enterprise data-policy routing, audit trail.

TASK

Compare OpenRouter PAYG, OpenRouter BYOK, and direct provider calls.

Pending.

↳ cost, setup time, observability, provider control, subprocessor complexity.

BENCHMARK COVERAGE · 1 BENCHMARK

BFCL ⚠

Benchmarks that evaluate or report on this tool. Scores and caveats live on the benchmark page.

PRIVACY DEEP-DIVE · checked 2026-06-09

OpenRouter routes prompts/completions to upstream model providers, so privacy depends on account settings, provider selection, provider logging/training policies, and Enterprise data-policy routing. Official docs/search snippets emphasize provider selection, model fallbacks, ZDR, provider logging, privacy best practices, input/output logging, and data-policy tags on model pages. OpenRouter docs indicate provider data-policy tags are best-knowledge indicators, not definitive third-party policies. Treat fallback routing as a privacy event: processor, logging, retention, latency, output quality, and cost can change unless constrained.

Training on your dataNot stated
EU data residencyEnterprise tier only
SOC 2 attestationNot documented
Local-first by defaultCloud only

POWERED BY

The model(s) that drive this card. The tool is the wrapper; the model is the engine.

ALTERNATIVES

LiteLLMTogether AIFireworks AIAmazon BedrockDirect OpenAI API

Compare side-by-side → open compare

WORKFLOW ROLES

How this tool fits into a composed research stack:

model routingAPI comparisoncost-sensitive model testing

QUALITY GATE · SOLID

Evidence ready
Benchmark pending
  • desk hands-on 2026-06-15: headline and metrics adjusted after desk trial.
  • score-revision 2026-06-09: differentiated from public facts; pending desk verification
  • Flagship-quality promotion 2026-06-04: source-backed dossier promoted; benchmark and public recommendation remain locked until evidence packet is complete.
  • Next action: Run six-model routing, fallback reliability, provider-data-policy, input/output logging, ZDR, and BYOK economics benchmarks.
  • Has API: true
  • Public recommendation: false
  • evidence-override: Notion source-of-truth sync; evidence remains source-backed until desk benchmark or hands-on pass.
  • Flagship-ready structure generated from Notion deep research and imported git fields on 2026-06-09.
Editorial · hands-onSolid

The Pack · Editorial newsletter

New cards in your inbox. Free.

One short email when a card ships or changes status. No tracking, no third-party analytics. Unsubscribe in one click.