Model catalog and prices
12 models, one endpoint: POST https://pay.omnidev.tech/v1/chat/completions. Prices are USD per million tokens and come straight from https://pay.omnidev.tech/v1/models.
Live data as of 2026-10-16 15:20 UTC. This page is generated from the live API (https://pay.omnidev.tech/v1/models and /v1/packages); the API itself is always authoritative.
| Model | Tier | Input / 1M tokens | Output / 1M tokens | Context (tokens) | Max output | Max output on MPP* | Pay with | Status |
|---|---|---|---|---|---|---|---|---|
kimi-k3 | Open-weight | $0.08 | $0.24 | 1,048,576 | 8,192 | 8,192 | MPP (Tempo), Prepaid package | Live (verified 2026-10-16) |
deepseek-v4.1-flash | Open-weight | $0.06 | $0.06 | 5,000,000 | 8,192 | 8,192 | MPP (Tempo), Prepaid package | Live (verified 2026-10-16) |
glm-5.3 | Open-weight | $0.40 | $0.20 | 5,000,000 | 8,192 | 8,192 | MPP (Tempo), Prepaid package | Live (verified 2026-10-16) |
llama-3.1-8b | Open-weight | $0.06 | $0.12 | 16,384 | 8,192 | 8,192 | MPP (Tempo), Prepaid package | Live (verified 2026-10-16) |
gpt-oss-120b | Open-weight | $0.12 | $0.50 | 122,880 | 8,192 | about 3,980 | MPP (Tempo), Prepaid package | Live (verified 2026-10-16) |
gpt-6-luna | Frontier lite | $0.15 | $0.75 | 1,000,000 | 8,192 | about 2,645 | MPP (Tempo), Prepaid package | Live (verified 2026-10-16) |
claude-haiku-5.5 | Frontier lite | $0.15 | $0.75 | 100,000 | 8,192 | about 2,645 | MPP (Tempo), Prepaid package | Live (verified 2026-10-16) |
gpt-6-sol | Frontier pro | $2.40 | $12.00 | 1,000,000 | 8,192 | about 145 | MPP (Tempo), Prepaid package | Live (verified 2026-10-16) |
claude-sonnet-5.5 | Frontier pro | $2.40 | $12.00 | 1,000,000 | 8,192 | about 145 | MPP (Tempo), Prepaid package | Live (verified 2026-10-16) |
gemini-3.1-pro | Frontier pro | $2.40 | $14.40 | 200,000 | 8,192 | about 120 | MPP (Tempo), Prepaid package | Live (verified 2026-10-16) |
grok-4.7 | Frontier pro | $2.40 | $7.20 | 500,000 | 8,192 | about 250 | MPP (Tempo), Prepaid package | Live (verified 2026-10-16) |
gpt-6-astra | Frontier max | $12.00 | $60.00 | 1,000,000 | 8,192 | about 15 | MPP (Tempo), Prepaid package | Listed; the $0.002 MPP request limit allows only very short outputs, so use a prepaid package when frontier credit opens |
* A single MPP request is limited to $0.002 in total (the quote is computed from your input size and max_tokens). The column shows the largest max_tokens that fits for a one-word prompt. If a request would exceed the limit you get 503 payment_rails_closed and nothing is charged; lower max_tokens or use a prepaid package.
What the status column means
- Live (verified date): a real paid request on that model completed end to end (402, payment, result) on that date.
- Listed, end-to-end check pending: the model is in the catalog and returns a correct price quote, but we have not yet completed a paid request on it. It may fail or be unavailable; a failed request is never charged.
- Listed, with a stated limit: the model works in principle but a current limit (for example the per-request MPP limit below) restricts it.
- Temporarily unavailable: the model is listed but currently failing checks.
How a request is priced
- MPP (Tempo): a fixed quote for the request: your input size and
max_tokensat the list price, with a minimum of $0.0008. Setmax_tokensto what you need. You pay only after the result is produced. - Prepaid packages: tokens are drawn down from a package balance; some models count as more than one token per token (the multiplier is in /v1/models as
package_token_multiplier). See Packages. - x402 upto (Base): pay the metered amount up to a signed maximum. Not enabled yet (coming soon); the fixed-price x402 routes below are live.
Fixed-price routes (x402 and MPP)
| Service | Routes | Price per call | Limits |
|---|---|---|---|
| Chat - nano | /v1/chat/nano (MPP)/x402/chat-nano (x402) | $0.001 | <= 1000 in / 256 out tokens |
| Chat - small | /v1/chat/small (MPP)/x402/chat-small (x402) | $0.002 | <= 8000 in / 1024 out tokens |
| Cited brief | /v1/brief (MPP)/x402/brief (x402) | $0.002 |
Limits and behaviour
- No streaming and no tool calls. Requests and responses follow the OpenAI chat-completions format.
- Models that reason before answering can spend the whole
max_tokensbudget on reasoning. If no text is produced the request fails with 502 and is not charged. Usemax_tokensof 256 or more for those models. - While the upstream is unavailable, paid routes answer 503 without charging you.
- Prices can change as supplier rates move; the 402 quote for your request is the price you pay.
Next: Quick start · API reference · Packages