Live state of the Venice catalog and token market, refreshed on every load.
Catalog by type
Privacy mode across all models
Private: Venice runs the model and keeps no prompt data. Anonymized: the request goes to a third-party provider with the user's identity removed. Field model_spec.privacy.
Service status
Added in the last 30 days
| Model | Type | Privacy | Added | Price |
|---|
Every model in GET /api/v1/models?type=all, with the fields an ops owner filters on.
| Model | Type | Privacy | Price (USD = DIEM) | Context | Flags |
|---|
Venice sells the same models two ways: direct on api.venice.ai, and as a provider inside OpenRouter. This tab matches each Venice text model to OpenRouter, then reads every provider's live price and 24h uptime for it.
| Model | Direct, blended $/1M | OpenRouter (Venice) | Cheapest provider | Venice rank | Uptime 24h | Status |
|---|
Blended price = input and output list prices weighted by the chosen input share, per 1M tokens. Rank counts every provider endpoint for the model, cheapest first. OpenRouter shows the price after any temporary provider discount and reports the discount; parity is judged on the list price. Uptime is OpenRouter's own 24h measure (rate limits and caller errors excluded); Venice's is compared with the median of the other providers. E2EE and fast-lane SKUs are separate products and are shown without a parity score. Click a row for every provider.
Resale pricing on first-party models
Venice's price for OpenAI, Anthropic, Google and xAI models against the vendor's own standard-tier endpoint on OpenRouter (tag openai, anthropic, google-ai-studio, xai), with any temporary discount divided out. OpenRouter passes vendor prices through without markup. The data cannot tell a vendor promotion from a standard price, so a ratio below 1.00 is a prompt to check the vendor's pricing page.
| Model | Privacy | Venice in / out | Vendor list in / out | Venice ÷ vendor |
|---|
Each staked DIEM carries $1 of API credit per day, and catalog prices are quoted 1:1 in USD and DIEM. Size a workload, see how many DIEM cover it, and compare holding DIEM with paying per token.
What one DIEM buys per day (75% input share)
| Model | Input $/1M | Output $/1M | Tokens per DIEM-day | Privacy |
|---|
Payback = DIEM capital divided by the USD you would otherwise spend per day. Holding DIEM is cheaper than paying per token when the daily cost of the capital tied up is below the daily bill; the DIEM itself keeps a market price, so the real risk is that price moving. Implied credit yield = 365 / DIEM price. DIEM staked and minted are read live from the DIEM contract on Base (totalStaked, totalSupply). Not financial advice.
Read from Venice's contracts on Base in one Multicall3 call: DIEM totalStaked and totalSupply, the staking contract's sVVV supply, totalLockedStakedVVV and mint curve, VVV supply and the burn balance at the zero address. Buybacks come from DefiLlama.
DIEM mint curve x: DIEM supply · y: sVVV locked per new DIEM
Points from diemSupply(i) and diemMintRates(i) on the staking contract. Shown from 25,000 to 45,000 DIEM; the curve keeps rising steeply after that. The marker is today's DIEM supply; the dashed line is the 40,000 target.
VVV bought back and burned daily, last 90 days
DefiLlama dailyHoldersRevenue for Venice: USD of VVV bought back. Offchain revenue is outside this series by design.
What changed in the catalog: against the snapshot bundled with this desk on 9 October 2026, and against your own last visit (kept in this browser only).
Since the 9 October 2026 baseline
Since your last visit
Rule-based checks over the live catalog, Venice's docs and the channel data. Each rule says why it matters and links to the evidence; a rule with no hits is listed as passing. Channel rules need the provider data from tab 3.
Do Venice's own sources agree?
The weekly brief this desk produces, written from the live data. Copy it into a doc or a channel.