Skip to content
WebmasterID

Transparency

Coverage

Verification state across the entity graph. This page reports exactly what has been verified against primary sources, what has been attempted but blocked, and what is still pending. There is no fabricated coverage here — if a field is unverified, this page says so.

At a glance

Entity graph summary

Counts derived from the typed local data layer. Cards link into the corresponding filtered hub view (filtered URLs are noindex, follow).

Freshness

Freshness and reverification queue

Coverage is not the same as freshness. A verified field can age into a review_due or stale state before any vendor change; the reverification queue points a human reviewer at the next source to re-check. The queue is informational — no automatic scraping, no auto-mutation.

Open reverification queue →

By provider

Per-provider verification

  • Anthropic (identification mark)
    Anthropic

    Last checked: 2026-05-20

    Verified
    Models verified
    4 / 4 tracked
    Verified pricing rows
    28

    No coverage notes recorded for this provider yet.

    Vendor-status observation API →

  • OpenAI (identification mark)
    OpenAI

    Last checked: not yet

    Partial
    Models verified
    0 / 1 tracked
    Verified pricing rows
    0

    URLs are the public documentation entry points. As of 2026-05-20 the platform.openai.com docs site returned HTTP 403 to automated retrieval; primary-source verification requires a manual browser review. No GPT-5 metrics have been published on this platform until that review completes.

  • Google (identification mark)
    Google

    Last checked: 2026-05-20

    Verified
    Models verified
    1 / 1 tracked
    Verified pricing rows
    11

    Per-model documentation, pricing reference, and API quickstart actively traversed on 2026-05-20 (see Gemini 2.5 Pro model record citations).

    Vendor-status observation API →

  • Meta (identification mark)
    Meta

    Last checked: 2026-05-21

    Verified
    Models verified
    2 / 2 tracked
    Verified pricing rows
    0

    Sprint 18 verification: Meta's Llama 4 model card at llama.com/docs/model-cards-and-prompt-formats/llama4 is publicly reachable and lists Scout (10M ctx) and Maverick (1M ctx) with input modalities, parameter counts, and knowledge cutoff. Because Meta does not host a paid API, per-model `pricing` rows are intentionally empty on the Llama 4 entries; max output and a stated release date are not on the model card and remain null.

  • Mistral (identification mark)
    Mistral

    Last checked: 2026-05-21

    Verified
    Models verified
    1 / 2 tracked
    Verified pricing rows
    2

    Models overview, models table, API reference, and the per-model spec card for Mistral Large 3 (v25.12) all retrievable. Sprint 16 verification pass landed the new spec-card URL pattern (/models/model-cards/<slug>), unlocking context window (256k) and pricing ($0.5/$1.5 per 1M input/output) on Mistral Large 3. Max output and explicit modality enumeration remain null because the spec card describes the model as 'multimodal' without listing channels.

  • DeepSeek (identification mark)
    DeepSeek

    Last checked: 2026-05-21

    Verified
    Models verified
    1 / 2 tracked
    Verified pricing rows
    3

    API docs root, chat-completion API reference, and Models & Pricing page were all retrieved on 2026-05-20 and re-retrieved on 2026-05-21 (Sprint 8B) — no value changed. Used to verify deepseek-v4-pro end-to-end (API string, lifecycle, context window, input cache-miss / cache-hit / output rates). The R1-0528 release announcement anchors the historical deepseek-r1 entry.

  • Groq (identification mark)
    Groq

    Last checked: 2026-05-21

    Verified
    Models verified
    0 / 0 tracked
    Verified pricing rows
    0

    Sprint 18 verification: Groq's supported-models page at console.groq.com/docs/models is publicly reachable and lists hosted third-party models with context windows and per-million-tokens pricing. The catalogue intentionally does NOT carry per-model entries under provider=groq because those models (Llama 3.x, GPT-OSS variants, Qwen, etc.) are not Groq-created. Groq verification is for the provider/platform itself.

  • Together AI (identification mark)
    Together AI

    Last checked: 2026-05-21

    Verified
    Models verified
    0 / 0 tracked
    Verified pricing rows
    0

    Sprint 18 verification: Together AI's serverless-models page at docs.together.ai/docs/serverless-models is publicly reachable and lists hosted third-party model families across chat, image, video, audio, and embedding categories. The catalogue intentionally does NOT carry per-model entries under provider=together-ai because those models are not Together-created. Together verification is for the provider/platform itself.

Status observation coverage

3 observers registered across the provider graph. Vendor-reported observers and independent HTTP probes are kept strictly separate; each observation records its own `source`.

  • Anthropic: vendor-status observer + independent HTTP probe. Live: /api/status/anthropic · /latest · /window?hours=24.
  • Google: vendor-status observer. Live: /api/status/google · /latest · /window?hours=24.
  • Uptime percentage: not computed. WebmasterID does not publish an uptime number without durable observations over a meaningful window — currently a minimum of 24 stored observations.
  • Cron: hourly via Vercel Cron against /api/cron/status; bearer-token-guarded in production.
  • Durable storage: configured. Observations are persisted via KV_REST_API_URL / KV_REST_API_TOKEN; when unset, the cron logs skipped_no_store and nothing is written.
  • Uptime window eligibility: not yet. Requires at least 24 stored observations in the window before any uptime-shaped number is exposed by /api/status/[provider]/window. The /status page itself does not display this number.

At a glance

Provider coverage matrix

Verified models, pricing rows, and status observers per provider. Computed from the local data layer at render time — no stale snapshots.

Coverage methodology →
Provider coverage by verified dimension
ProviderStatusVerified modelsVerified pricing rowsAPI docsPricing docsStatus observer
Anthropicverified4 / 428linkedlinked2 observers
OpenAIpartial0 / 10linkedlinked—
Googleverified1 / 111linkedlinked1 observer
Metaverified2 / 20———
Mistralverified1 / 22linkedlinked—
DeepSeekverified1 / 23linkedlinked—
Groqverified0 / 00linkedlinked—
Together AIverified0 / 00linkedlinked—

Sprint 8B — DeepSeek & Mistral verification summary

  • DeepSeek V4 Pro: API string, lifecycle, context window, input cache-miss, input cache-hit, and output pricing verified end-to-end against DeepSeek's Models & Pricing page and chat-completions API reference. Re-verified on 2026-05-21 — values unchanged. The 75% promotional discount on deepseek-v4-pro is still active until 2026/05/31 15:59 UTC; the regular rate is recorded as the durable value.
  • DeepSeek R1 (historical): kept as a historical record anchored to the R1-0528 release announcement. The R1 family is no longer in the current API model parameter list, so the entry is recorded as retired with deepseek-v4-pro as the documented migration target.
  • Mistral Large 3: API string and lifecycle verified from Mistral's models overview and table. Per-model spec card pages still 404 to automated retrieval; the API pricing tab on mistral.ai/pricing is still JS-driven. Pricing, context window, max output, and modality remain unverified pending a manual browser pass.
  • Mistral Large 2: Sprint 8B's preferred target is documented in Mistral's Legacy/Deprecated table as deprecated 2024-11-30 and retired 2025-03-30 — already retired today. The catalogue carries it as a historical entry; new workloads should evaluate Mistral Large 3.
  • Gemini cache pricing: the schema keeps Google's per-hour cache storage rate (1M cache storage / hour) distinct from Anthropic's TTL-write rates. We do not map one provider's cache semantics into another's.

Audit log

Primary-source retrieval attempts

Every URL targeted during a verification pass, the outcome, and a free-text note. This is what makes the platform auditable.

ProviderTargetURLAttemptedResult
AnthropicClaude family model specificationsplatform.claude.com/docs/en/docs/about-claude/models/overview2026-05-20Verified
AnthropicClaude pricing referenceplatform.claude.com/docs/en/about-claude/pricing2026-05-20Verified
GoogleGemini 2.5 Pro model specificationsai.google.dev/gemini-api/docs/models/gemini-2.5-pro2026-05-20Verified
GoogleGemini pricing referenceai.google.dev/pricing2026-05-20Verified
GoogleGemini API request shapeai.google.dev/gemini-api/docs/quickstart2026-05-20Verified
OpenAIGPT-5 model specificationsplatform.openai.com/docs/models/gpt-52026-05-20Blocked (403)
OpenAIOpenAI pricing referenceplatform.openai.com/docs/pricing2026-05-20Blocked (403)
OpenAIOpenAI model catalogueplatform.openai.com/docs/models2026-05-20Blocked (403)
OpenAIOpenAI marketing-side pricingopenai.com/api/pricing2026-05-20Blocked (403)
AnthropicAnthropic brand resource pagewww.anthropic.com/brand2026-05-20404
DeepSeekDeepSeek API docs root (active vs deprecated models)api-docs.deepseek.com2026-05-20Verified
DeepSeekDeepSeek Models & Pricingapi-docs.deepseek.com/quick_start/pricing2026-05-20Verified
DeepSeekDeepSeek chat completion API referenceapi-docs.deepseek.com/api/create-chat-completion2026-05-20Verified
MistralMistral models overviewdocs.mistral.ai/getting-started/models/models_overview2026-05-20Verified
MistralMistral models table (API strings)docs.mistral.ai/getting-started/models2026-05-20Verified
MistralMistral chat completions API referencedocs.mistral.ai/api2026-05-20Verified
MistralMistral Large 3 per-model spec carddocs.mistral.ai/getting-started/models/mistral-large-32026-05-20404
MistralMistral API pricingmistral.ai/pricing2026-05-20Manual browser only
DeepSeekDeepSeek Models & Pricing (Sprint 8B re-verification)api-docs.deepseek.com/quick_start/pricing2026-05-21Verified
DeepSeekDeepSeek chat completion API reference (Sprint 8B re-verification)api-docs.deepseek.com/api/create-chat-completion2026-05-21Verified
DeepSeekDeepSeek R1-0528 release announcement (historical R1 anchor)api-docs.deepseek.com/news/news2505282026-05-21Reviewed
MistralMistral Large 2 lifecycle confirmation (Sprint 8B)docs.mistral.ai/getting-started/models/models_overview2026-05-21Verified
MistralMistral Large 3 per-model spec card (Sprint 8B retry)docs.mistral.ai/getting-started/models/mistral-large-32026-05-21404
MistralMistral API pricing (Sprint 8B retry)mistral.ai/pricing2026-05-21Manual browser only
MistralMistral Large 3 (v25.12) per-model spec carddocs.mistral.ai/models/model-cards/mistral-large-3-25-122026-05-21Verified
OpenAIOpenAI GPT-5 model docs (Sprint 16 retry)platform.openai.com/docs/models/gpt-52026-05-21Blocked (403)
MetaMeta Llama 4 model card (Scout + Maverick)www.llama.com/docs/model-cards-and-prompt-formats/llama42026-05-21Verified
GroqGroq supported-models referenceconsole.groq.com/docs/models2026-05-21Verified
Together AITogether AI serverless-models referencedocs.together.ai/docs/serverless-models2026-05-21Verified
OpenAIOpenAI GPT-5 model docs (Sprint 18 retry)platform.openai.com/docs/models/gpt-52026-05-21Blocked (403)
GroqGroq hosted-model pricing tablegroq.com/pricing2026-05-23Verified
Together AITogether AI serverless inference pricingwww.together.ai/pricing2026-05-23Verified
Together AITogether AI serverless-models docs (Llama 4 lookup)docs.together.ai/docs/serverless-models2026-05-23insufficient-data
Together AITogether AI pricing page (Llama 4 Scout/Maverick rows)www.together.ai/pricing2026-05-23ambiguous-model-mapping

What you will not find on this site

  • Benchmark leaderboards with scores. Scores are not republished without an independent primary-source reference.
  • Estimated latency or uptime numbers. Data not yet verified. renders in their place until an independent monitor is wired.
  • Vendor-provided pricing where the vendor's documentation page could not be retrieved. The blocked entries above record exactly which pages.