TokenDos Public Evidence Snapshot

Prices and statistics are supplied by the TokenDos operator and are not an independent third-party test. Page rendered at (UTC): 2026-11-02T12:56:35.170309921Z

Availability data is cached for 5 minutes and HTML responses may be cached for 10 minutes; these delays can combine. Prices, listings and statistics are queried separately, not as an atomic snapshot. Render time does not verify price freshness.

Snapshot summary

All current marketplace models and prices

Every model returned by this marketplace query is included, even without an official match. Site input/output minima convert marketplace prices per 1K tokens ×1000 to USD/1M tokens; the two minima may come from different nodes. Mixed-billing and non-TOKEN models do not show token prices or rates. Per-call/per-second prices are published minima; see marketplace details for resolution and other conditions. Official references are operator-maintained prices per 1M tokens in their original currency. Published platform rate reuses public terms: input rate = (site USD input price × CNY/USD FX) / official catalog input price, with the actual FX shown below. This formula does not adjust for catalog currency: for USD references it compares numbers in different currencies and cannot be read as a discount, savings percentage, or 1x price parity. Missing official prices, site token prices or FX leave rates unreported.

Reference response date is the UTC date assigned by the public terms service, not the last verification date of upstream official prices. Catalog entries may be missing or stale; this date does not establish price freshness.

Success-rate samples and scale methodology

Statistics aggregate API_CALL_LOG by day. successCalls counts status=SUCCESS; success rate is successCalls/totalCalls and reflects recorded HTTP-layer request outcomes, not independent uptime monitoring, answer quality or an SLA. The current UTC day is unfinished, so the final day is partial. With no samples, success rate is unreported, not 0%.

The service normally filters logs to currently listed models; an empty or failed marketplace lookup may retain unfiltered historical logs in the public API, marked by modelScope. queryStatus=ok means a successful query with samples, no_samples a successful query without samples, and unavailable a failed aggregate query. This page shows samples and success rates only after a successful query with modelScope=current_marketplace; other states hide these figures. Degraded results are not written to the availability service cache, but HTML responses may still be cached for 10 minutes. The reference service can degrade internally without a status flag; an absent match does not prove no official price exists. Active nodes include active/throttled providers and exclude suspended seats; this service-wide count is not a count of purchasable marketplace nodes.

Public sources and limitations

• Current marketplace listings and minima: https://www.tokendos.com/api/tokendos/provider/marketplace
• Official references, rates and terms: https://www.tokendos.com/api/tokendos/public/terms
• Success-rate aggregate and samples: https://www.tokendos.com/api/tokendos/public/availability
• Collector-compatible snapshot (generated separately): https://www.tokendos.com/api/tokendos/public/transit-snapshot
• Full machine-readable description: /llms-full.txt; listing reference: /tokendos-listing
• Data is maintained by the TokenDos operator. Upstream pools, sampling coverage, network paths and windows affect results. Cite page render time, statistics generation time, window dates and sample counts together.

Related pages: /tokendos-pricing · /tokendos-market