Skip to content
WebmasterID

Provider

Groq

Inference infrastructure provider. Hosts third-party open-weights models (Llama, GPT-OSS, Qwen, Whisper, etc.) on custom LPU hardware — Groq is a hosting platform, not a model creator.

Groq (identification mark)

Groq

Verified

Mountain View, USA

Last checked: 2026-05-21 · brand mark: nominative

Official sources

Primary links

Vendor-published documentation roots. Linked content has not been individually verified as primary-source citations unless this provider's verification status is 'verified'.

Catalogue

Tracked models

0 models in the catalogue, 0 verified end-to-end.

No Groq models in the catalogue yet.

Hosted platform

Groq hosts third-party models

Groq is a hosting/inference platform — it bills for models created by other organisations. The rows below are pricing references sourced from Groq's own pricing page, not live quotes. Pricing is volatile and may change after retrieval. Treat prices as source-backed references, not live quotes.

ModelModel creatorHosted model IDInput / 1MOutput / 1MFreshnessSource
Llama 4 ScoutMetameta-llama/llama-4-scout-17b-16e-instruct$0.11$0.34FreshGroq

Hosted pricing is set by Groq — not by the model's creator. Rows are source-backed references, not live quotes; WebmasterID Models does not rank hosting platforms by price. See methodology for the full distinction.

What we cover

Data coverage

Which areas have verified data for this provider on the WebmasterID Models graph.

AreaCoverage
PricingNo verified pricing yet
Models0 tracked, 0 verified
BenchmarksNo verified third-party benchmark scores yet
InfrastructureNo verified region/latency/uptime data yet
StatusNo independent monitor wired yet

Data gaps

Fields not yet verified

These fields are intentionally null until verified against a primary source. They are not estimated.

  • Deprecations URL

    Provider does not publish a stable deprecations page in their docs, or it has not yet been recorded.

  • Status observer

    No vendor-status or independent-probe observer wired for this provider yet.

  • Regional availability

    Per-provider regions array is null across the catalogue. Wiring Bedrock / Vertex availability matrices is a planned data expansion.

  • Per-account rate-limit numbers

    Rate-limit tiers live in account consoles and are not surfaced as verified fields on this site.

How a value moves from this list to a verified row — see source verification methodology.

Entity graph

Related coverage

Continue

Related providers

Providers that appear alongside this one in comparisons, plus the rest of the catalogue.

Audit log

Verification attempts

Every primary-source URL targeted for this provider and what came back. Blocked attempts are recorded honestly rather than substituted with second-hand data.

All coverage →
  • Groq supported-models reference

    console.groq.com/docs/models

    Groq supported-models page retrieved. Lists hosted third-party models (Llama 3.x, GPT-OSS, Qwen, Whisper, etc.) with context windows and pricing. Used only to verify the provider/platform itself — these models are NOT Groq-created, so no per-model entries are added under provider=groq.

    2026-05-21

    verified

  • Groq hosted-model pricing table

    groq.com/pricing

    Groq pricing page retrieved as raw HTML. The serverless inference table cleanly anchors 'Llama 4 Scout (17Bx16E) 128k' at $0.11/1M input and $0.34/1M output — recorded as a hosted-provider pricing row (creator=meta, billing=groq). Llama 4 Maverick is NOT on the Groq pricing table; no row added for Maverick on Groq. Page also documents a 50% cache-hit discount and a Batch API (50% off, 24h–7d window); those are noted on the hosted row but not synthesised into derived prices.

    2026-05-23

    verified