# How much does LLM API cost per token?

Each LLM API model is billed per million input and output tokens at 5% below its list price. Rates run from DeepSeek V4 Flash input, at $0.1805 per million tokens, to Claude Opus 5 output, at $23.75 per million tokens. Every call is billed by exact token usage and drawn from one prepaid credit balance, which you buy from $0.50 with a bearer credential, the agent identity the storefront issues.

## At a glance

- **Endpoints:** `POST https://agents.llmapi.ai/zcj/76dh2uh0pe2s/zc/v1/chat/completions`, `/zc/v1/messages`, `/zc/v1/responses` and `/zc/v1/embeddings`. Set `"stream": false`, because the storefront does not stream.
- **Price:** per million input and output tokens, 5% below list price, by exact usage. No tokens are included.
- **Credit:** the Prepaid credit plan, from $0.50 per purchase and $0.50 by card. Top up with `POST https://agents.llmapi.ai/zcj/76dh2uh0pe2s/extend`.
- **Payment:** x402 (Base USDC), MPP (Tempo USDC) or card. Every purchase needs a bearer credential.
- **Token counting:** `POST https://agents.llmapi.ai/zcj/76dh2uh0pe2s/zc/v1/messages/count_tokens`, free ($0 per request).
- **Cost per call:** each response's `zc-billing` header gives `chargedUsd` and `remainingCreditUsd`.
- **Full purchase flow:** [the buyer skill](https://agents.llmapi.ai/zcj/76dh2uh0pe2s/SKILL.md)

## LLM API price per million tokens, by model

| Model | `model` value | Context | Input | Output |
| --- | --- | --- | --- | --- |
| Claude Opus 5 | `claude-opus-5` | 1M | $4.75 per million tokens | $23.75 per million tokens |
| Claude Sonnet 5 | `claude-sonnet-5` | 1M | $1.90 per million tokens | $9.50 per million tokens |
| GPT-5.2 | `gpt-5.2` | 400K | $1.6625 per million tokens | $13.30 per million tokens |
| GPT-5.1 | `gpt-5.1` | 400K | $1.1875 per million tokens | $9.50 per million tokens |
| GPT-5 Mini | `gpt-5-mini` | 400K | $0.2375 per million tokens | $1.90 per million tokens |
| Gemini 3.1 Pro (Preview) | `gemini-3.1-pro-preview` | 1M | $1.90 per million tokens | $11.40 per million tokens |
| Gemini 3.7 Flash | `gemini-3.7-flash` | 1M | $0.7125 per million tokens | $3.5625 per million tokens |
| DeepSeek V4 Pro | `deepseek-v4-pro` | 1M | $1.653 per million tokens | $3.306 per million tokens |
| DeepSeek V4 Flash | `deepseek-v4-flash` | 1M | $0.1805 per million tokens | $0.4845 per million tokens |
| Grok 4.6 | `grok-4.6` | 500K | $1.90 per million tokens | $5.70 per million tokens |
| Kimi K2.6 | `kimi-k2.6` | 262K | $0.9025 per million tokens | $3.80 per million tokens |
| GLM 5.3 | `glm-5.3` | 1M | $1.33 per million tokens | $4.18 per million tokens |
| Text Embedding 3 Large | `text-embedding-3-large` | 8K per input | $0.1235 per million tokens | Input only |
| Text Embedding 3 Small | `text-embedding-3-small` | 8K per input | $0.019 per million tokens | Input only |
| Jina Embeddings v4 | `jina-embeddings-v4` | 32K per input | $0.019 per million tokens | Input only |

Send the `model` value exactly as written. A service slug such as `gpt-5-1` is not a valid value.

## Estimate a call's cost before sending it

1. Count the input tokens. Use the same `model` and `messages` you plan to send:

   ```http
   POST https://agents.llmapi.ai/zcj/76dh2uh0pe2s/zc/v1/messages/count_tokens
   Authorization: Bearer <access token>
   Content-Type: application/json

   {"model": "claude-sonnet-5", "messages": [{"role": "user", "content": "In one sentence, what does an HTTP 402 Payment Required response mean?"}]}
   ```

   The response returns the count:

   ```json
   {"input_tokens": 22, "model": "claude-sonnet-5"}
   ```

   The count drew nothing from prepaid credit.
2. Multiply the count by the model's input rate. At that count, 1,000 such requests on Claude Sonnet 5 cost $0.0418 in input.
3. Cap output with `max_tokens`, or with `max_output_tokens` on `/zc/v1/responses`. Output is billed on the tokens the model returns, and the cap sets how many it can return.

Worked costs, taking a short chat call as 200 input and 100 output tokens:

| Job | Model | Cost |
| --- | --- | --- |
| 1,000 short chat calls | DeepSeek V4 Flash, the lowest rate | $0.08455 |
| 1,000 short chat calls | Claude Opus 5, the highest rate | $3.325 |
| Embed 1M tokens | Text Embedding 3 Small | $0.019 |
| Embed 1M tokens | Text Embedding 3 Large | $0.1235 |

## Buy and top up prepaid credit

- Buy credit once with your bearer credential, from $0.50 by wallet or $0.50 by card, and top up the same balance later with `POST https://agents.llmapi.ai/zcj/76dh2uh0pe2s/extend`. [How to pay](https://agents.llmapi.ai/zcj/76dh2uh0pe2s/payment.md) covers each rail.
- Calls draw down the balance: each deducts its exact token cost, and unused credit stays on it.
- A call without a credential answers `401 bearer_required`. A credential without credit answers `403 plan_required`. A spent balance answers `usage_exhausted`, and a top-up clears it.
- The [buyer skill](https://agents.llmapi.ai/zcj/76dh2uh0pe2s/SKILL.md) has the steps for registering the credential.

## Frequently asked questions

### What does Claude Opus 5 cost per token?

Claude Opus 5 costs $4.75 per million tokens for input and $23.75 per million tokens for output, 5% below list price, with a 1M context. Send `"model": "claude-opus-5"` to chat completions, messages or responses.

### What does GPT-5 Mini cost per token?

GPT-5 Mini costs $0.2375 per million tokens for input and $1.90 per million tokens for output, with a 400K context. Use `"model": "gpt-5-mini"`. At 200 input and 100 output tokens per call, 1,000 calls cost $0.2375.

### Which LLM API model is cheapest per token?

Among chat models, DeepSeek V4 Flash has the lowest rates on both meters: $0.1805 per million tokens for input and $0.4845 per million tokens for output. For embeddings, Text Embedding 3 Small and Jina Embeddings v4 both cost $0.019 per million tokens.

### Are there free tokens or a free trial?

No. Every meter lists zero included units, and the catalog lists no trial. Token counting is free ($0 per request), but the request still needs a bearer credential that holds Prepaid credit.

### Can a call be paid on its own, without buying credit?

No. Every call draws on prepaid credit held by a bearer credential. To start, buy at least $0.50 of credit with that credential, then send calls with `Authorization: Bearer <access token>`.

## Keep reading

- [Buyer skill](https://agents.llmapi.ai/zcj/76dh2uh0pe2s/SKILL.md): registering the credential and buying credit
- [How to pay](https://agents.llmapi.ai/zcj/76dh2uh0pe2s/payment.md): x402, MPP and card
- [Catalog](https://agents.llmapi.ai/zcj/76dh2uh0pe2s/manifest.json): live services, plans and prices
- [Error reference](https://agents.llmapi.ai/zcj/76dh2uh0pe2s/errors.md): every code and what to do

---

Updated 2026-10-28. All LLM API guides: https://agents.llmapi.ai/guides.md. Storefront: https://agents.llmapi.ai/zcj/76dh2uh0pe2s/llms.txt.