openDashboard MODEL CATALOGUE
npm latest
Meet the Model Router ↘
YOUR AGENT. MORE POSSIBILITIES.

Put the pitch to the test.

Find model entries and exact provider slugs, with published prices wherever available. Bring the catalogue to the agent you already use.

  • Choose with prices. Read published rates across providers, with their units and conditions.
  • Start with free. Surface OpenRouter :free entries so your agent can pick zero-cost options, subject to provider limits.
  • Go beyond text. Discover image and video generation entries alongside language models.
Install from npm

Connect model and price coverage to your MCP agent.

npx -y open-dashboard-mcp
Explore the catalogue

Entries include variants and pricing configurations. Some are not API-ready. A catalogue listing does not guarantee a successful call.

THE CATALOGUE, COUNTEDLIVE READ
Reading all registered providers…

More ways to build.

Fresh catalogue reads. Actual coverage.

Model entries
—

Distinct provider + ID pairs across — registered providers.

Entries with prices
—

Across — providers with prices.

OpenRouter free entries
—

Exact slugs ending in :free.

Image entries
—

— with native output prices.

Video entries
—

— with native output prices.

Counts will describe the entries returned by this live read.

Counts include variants and pricing configurations. Prices can be derived from published coefficients or depend on conditions. Native output rates retain image, video, duration, pixel or credit units; token-only rates do not count as native output prices. Unknown prices stay unknown.

Live read timestamp pending.

Provider counts and read status
ProviderModel entriesPriced entriesRead status
Inspect the live API ↗
What model or provider should I add next?

Opens a prefilled issue in the public SD Forest repository. Submit on GitHub to send your suggestion; sign-in is required and issues are public.

AUDIO. PUBLIC EVIDENCE.

Read the voice receipt.

Published provider prices and public voice-quality evidence, kept in their native units.

Read-only. No inference calls.
Unknown stays unknown.

THE EVIDENCE FILEElevenLabs / Cartesia
Reading provider sources…

Read the receipt.

Current source observations, with the fine print.

Samples, not a like-for-like ranking. Match model, settings and billing unit before comparing. Prices are native provider units; MOS is a dated public benchmark, not an Open Dashboard synthesis.

Reading the public pricing sources. Missing values will stay unknown.

Read the public TTS benchmark ↗Source read pending
READ THE UNIT. FOLLOW THE SOURCE.
THE PROVIDER FIELD

OpenRouter · Groq · Cerebras · Sail · Nous Research · QwenCloud · DeepInfra · Novita · SambaNova · Chutes · WaveSpeedAI · fal · Crazyrouter · AkashML · io.net · KIE

Fresh counts and provider read status appear above. Catalogue coverage can change.
— explorer entries (live + snapshots)— npm adapters— optional MCP tools Checking npm downloads… Reading the latest complete weekly window from npm. Verify on npm ↗ Checking npm releases… Reading published package versions from npm. Verify releases on npm ↗ Fresh counts and provider read status appear above. Catalogue coverage can change.

Let your agent compare

Prices worth comparing.

Listed-price examples checked . Quotes and account conditions can change.

GPT-4o mini · Text

Crazyrouter: 35% lower listed token rates.

$0.0975 input / $0.39 output per million uncached tokens, versus OpenRouter’s $0.15 / $0.60.

Crazyrouter rates are derived from published coefficients and its 0.65 discount. The same model alias does not establish identical deployments or bills.

Crazyrouter source OpenRouter source

Seedream 4.5 · Image

KIE: 18.75% lower listed image price.

$0.0325 per image on KIE, versus $0.04 on fal.ai, for text-to-image with single-image 2K settings.

KIE: 6.5 credits at the standard $0.005 credit rate, excluding bonuses. The fal.ai quote comes from its official model page.

KIE source fal.ai source

Groq · Fast inference

Start on a limited Free tier.

Try fast inference with Groq’s Free tier. Limits apply per organization and model; check your account’s current quota.

Higher limits are available on Groq’s pay-as-you-go Developer plan.

Free-tier limits Plan details

THE MODEL ATLAS

Find the entry that fits.

Find a price, follow a source, inspect a possibility.
Live reads and dated snapshots keep their own labels.

Loading data
Compare two providers Choose two providers to compare their catalogues side by side.

A landscape of possibilities

Published token prices · USD per million tokens

Click or tap to inspect

Loading published model data…

Free comes with a few details.

Provider limits are shown with their source and checked date.

Groq: model-specific request and token limits.

Where this data comes from

Follow the connections

See the models behind the apps.

Real public usage, connected.
Trace a model to the tools that use it.

Checking common observation date…

Select an app, model or connection to inspect its evidence.

OpenRouter public observations only. Link width uses a square-root scale of observed tokens for one shared day. Missing links mean “not observed”, not zero usage or lack of compatibility.

From exploring to doing

All this information.
One conversation.

Your agent can look up model IDs, compare prices and check
what changed. Explore free options, image generation and video generation alongside language models. Connect the tools you need, and keep exploring.

Connect your agent ↗
HermesCodexClaude CodeCursorAny compatible MCP client

Read-only intelligence. Model inference still uses your chosen provider.

There’s more in the ecosystem

What’s trending on GitHub?

Browse trending repositories and curated AI projects.
Find Ivan’s GitHub profile and 3D repository library on the projects page.

Explore GitHub trending & projects ↗

A small part of a larger forest

Made in SD Forest.

A home for useful tools and curious experiments.
Take a look around whenever you like.

Visit SD Forest ↗
THE CONNECTED TOOLKIT

Evidence in.
A decision
you can read.

Model Router is an OpenAI-compatible proxy that compares measured candidates per request, records why, and abstains when no model qualifies.

The live read above measures catalogue breadth across providers, including OpenRouter.

Entries include variants and pricing configurations. Catalogue entries are not a count of qualified router candidates.

Model Router repository

A separate project. This dashboard reads evidence; it does not route your requests.

MODEL ROUTERANATOMY OF A DECISION

Start with what is known.

The router gathers current catalogue prices, measured model behavior and provider quota or budget state.

Illustrative routing flow. No request executed.

What “free” means here