# LLM API models: model ids, context sizes and per-token prices

This page lists every chat and embedding model on LLM API's agent storefront, with its `model` id, context size, price per million input and output tokens and the endpoints it answers on. DeepSeek V4 Flash has the lowest chat rates and a 1M-token context: input $0.1805 per million tokens; output $0.4845 per million tokens. Every model draws on the same Prepaid credit balance, so you switch models by changing the `model` field.

## At a glance

- **Models:** 12 chat models and 3 embedding models, in the tables below.
- **Chat endpoints:** `POST https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/zc/v1/chat/completions`, `POST https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/zc/v1/messages` and `POST https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/zc/v1/responses`. Every chat model answers on all three.
- **Embeddings endpoint:** `POST https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/zc/v1/embeddings`
- **Token counting:** `POST https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/zc/v1/messages/count_tokens` counts a message's input tokens before you send it, at $0.
- **Price:** billed per exact input and output token, at a rate per million tokens. Each rate is 5% off the model's list price. The catalog lists no free tokens on any meter.
- **Payment:** calls draw on Prepaid credit, from $0.50 per purchase ($0.50 by card), bought over x402, MPP or card with a bearer credential.
- **Streaming:** not supported. Set `stream` to `false` or omit it.
- **Switching models:** change `model`. The endpoint, credential and balance stay the same.
- **Live prices and full flow:** [manifest.json](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/manifest.json) and the [buyer skill](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/SKILL.md).

## Lowest-priced chat models: DeepSeek V4 Flash and GPT-5 Mini

The chat tables group models by output price. Rows run from the lowest output price to the highest.

| Model | `model` | Context | Input | Output |
| --- | --- | --- | --- | --- |
| DeepSeek V4 Flash | `deepseek-v4-flash` | 1M | $0.1805 per million tokens | $0.4845 per million tokens |
| GPT-5 Mini | `gpt-5-mini` | 400K | $0.2375 per million tokens | $1.90 per million tokens |

## Mid-priced chat models: DeepSeek V4 Pro, Gemini 3.7 Flash, Kimi, GLM and Grok

| Model | `model` | Context | Input | Output |
| --- | --- | --- | --- | --- |
| DeepSeek V4 Pro | `deepseek-v4-pro` | 1M | $1.653 per million tokens | $3.306 per million tokens |
| Gemini 3.7 Flash | `gemini-3.7-flash` | 1M | $0.7125 per million tokens | $3.5625 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 |
| Grok 4.6 | `grok-4.6` | 500K | $1.90 per million tokens | $5.70 per million tokens |

## Higher-priced chat models: GPT-5.1, GPT-5.2, Gemini 3.1 Pro and Claude

| Model | `model` | Context | Input | Output |
| --- | --- | --- | --- | --- |
| GPT-5.1 | `gpt-5.1` | 400K | $1.1875 per million tokens | $9.50 per million tokens |
| Claude Sonnet 5 | `claude-sonnet-5` | 1M | $1.90 per million tokens | $9.50 per million tokens |
| Gemini 3.1 Pro (Preview) | `gemini-3.1-pro-preview` | 1M | $1.90 per million tokens | $11.40 per million tokens |
| GPT-5.2 | `gpt-5.2` | 400K | $1.6625 per million tokens | $13.30 per million tokens |
| Claude Opus 5 | `claude-opus-5` | 1M | $4.75 per million tokens | $23.75 per million tokens |

Send the `model` value exactly as the tables write it. The catalog and price reports label services with slugs such as `gpt-5-2` or `gemini-3-1-pro-preview`, and a slug is not a valid `model` value.

## Embedding model prices

Embeddings bill input tokens only.

| Model | `model` | Context | Input |
| --- | --- | --- | --- |
| Text Embedding 3 Small | `text-embedding-3-small` | 8K per input | $0.019 per million tokens |
| Jina Embeddings v4 | `jina-embeddings-v4` | 32K per input | $0.019 per million tokens |
| Text Embedding 3 Large | `text-embedding-3-large` | 8K per input | $0.1235 per million tokens |

Embedding models answer only on `/zc/v1/embeddings`. Chat models do not answer there.

## Chat models by context size

Within each row, models run from the lowest output price to the highest.

| Context | Models |
| --- | --- |
| 1M | DeepSeek V4 Flash, DeepSeek V4 Pro, Gemini 3.7 Flash, GLM 5.3, Claude Sonnet 5, Gemini 3.1 Pro (Preview), Claude Opus 5 |
| 500K | Grok 4.6 |
| 400K | GPT-5 Mini, GPT-5.1, GPT-5.2 |
| 262K | Kimi K2.6 |

## Endpoints and request fields

| Endpoint | Request shape | Required fields | Output cap |
| --- | --- | --- | --- |
| `POST https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/zc/v1/chat/completions` | OpenAI-style chat completions | `model`, `messages` | `max_tokens` (optional) |
| `POST https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/zc/v1/messages` | Anthropic Messages | `model`, `messages`, `max_tokens` | `max_tokens` (required) |
| `POST https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/zc/v1/responses` | OpenAI Responses API, synchronous only | `model`, `input` | `max_output_tokens` (optional) |
| `POST https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/zc/v1/embeddings` | Embeddings | `model`, `input` (a string or an array of strings) | None |

An example chat request:

```bash
curl -sS -X POST https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/zc/v1/chat/completions \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer <token>' \
  -d '{"model": "deepseek-v4-flash", "stream": false, "messages": [{"role": "user", "content": "List three uses for a paper clip."}], "max_tokens": 200}'
```

To run the same request on another model, change only `"model"`, for example to `"claude-sonnet-5"`. The response is the upstream provider's status and JSON body, returned in one piece. A response funded from credit carries a `zc-billing` header with `fundingSource`, `chargedUsd` (what this call drew) and `remainingCreditUsd`. A delivered response with no `zc-billing` header billed nothing.

Boundaries: streaming requests and Responses API background mode are rejected. Each call bills the exact tokens it used. The output cap field bounds how many output tokens one call can bill.

## Worked costs per model

The totals are for a job of 100,000 input tokens and 10,000 output tokens, at today's rates.

| Model | Job total |
| --- | --- |
| DeepSeek V4 Flash | $0.022895 |
| GPT-5 Mini | $0.04275 |
| Gemini 3.7 Flash | $0.106875 |
| Kimi K2.6 | $0.12825 |
| GLM 5.3 | $0.1748 |
| DeepSeek V4 Pro | $0.19836 |
| GPT-5.1 | $0.21375 |
| Grok 4.6 | $0.247 |
| Claude Sonnet 5 | $0.285 |
| GPT-5.2 | $0.29925 |
| Gemini 3.1 Pro (Preview) | $0.304 |
| Claude Opus 5 | $0.7125 |

Embedding 10 million tokens costs $0.19 on Text Embedding 3 Small, $0.19 on Jina Embeddings v4 and $1.235 on Text Embedding 3 Large.

## Paying for model calls

Every model on this page draws on Prepaid credit, held by a bearer credential, the agent identity the storefront issues. Credit starts at $0.50 per purchase by wallet and $0.50 by card, paid over x402, MPP or card. A credit purchase answers with a 402, the price quote, which you pay and retry. Each call then deducts its exact token cost from the balance, and unused credit stays on it. The [buyer skill](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/SKILL.md) holds the full purchase flow, and [How to pay](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/payment.md) covers each rail.

## Errors when calling a model

| Code | Meaning | Action |
| --- | --- | --- |
| `401 bearer_required` | The call carried no agent credential | Register at [auth.md](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/auth.md), then retry with `Authorization: Bearer <token>` |
| `401 invalid_token` | The credential did not verify, for example `token_expired` | Exchange the stored assertion for a fresh token |
| `403 plan_required` | No plan on the credential covers the call | Buy Prepaid credit from the `plans` in the body, then retry |
| `usage_exhausted` | The credit balance ran out | Top up with `POST https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/extend` |
| `amount_below_minimum` | The purchase amount is under the plan minimum | Send at least $0.50 |

The full list is at [errors.md](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/errors.md).

## Frequently asked questions

### What is the cheapest model on LLM API?

DeepSeek V4 Flash has the lowest chat rates: input $0.1805 per million tokens; output $0.4845 per million tokens. It has a 1M-token context. DeepSeek V4 Pro, `deepseek-v4-pro`, costs $1.653 per million tokens for input and $3.306 per million tokens for output, also with a 1M context. For embeddings, Text Embedding 3 Small and Jina Embeddings v4 share the lowest input rate.

### How do Claude Opus 5 and Claude Sonnet 5 prices compare?

Both have a 1M-token context and answer on the same three chat endpoints. Claude Opus 5 costs $4.75 per million tokens for input and $23.75 per million tokens for output. Claude Sonnet 5 costs $1.90 per million tokens for input and $9.50 per million tokens for output. For a job of 100,000 input and 10,000 output tokens, that is $0.7125 on Opus 5 and $0.285 on Sonnet 5.

### What do GPT-5.2 and Gemini 3.1 Pro cost per token?

GPT-5.2, `gpt-5.2`, has a 400K context and costs $1.6625 per million tokens for input and $13.30 per million tokens for output. Gemini 3.1 Pro (Preview), `gemini-3.1-pro-preview`, has a 1M context and costs $1.90 per million tokens for input and $11.40 per million tokens for output.

### Can I switch models without changing anything else?

Yes, for chat models. All 12 answer on `/zc/v1/chat/completions`, `/zc/v1/messages` and `/zc/v1/responses` and draw on the same Prepaid credit balance. Change the `model` value and keep the rest of the request. Embedding models switch the same way, on `/zc/v1/embeddings` only.

### Do I need to buy credit before the first call?

Yes. Every model call draws on Prepaid credit held by a bearer credential. A call without a credential answers `401 bearer_required`, and a credential without credit answers `403 plan_required`. Credit starts at $0.50 by wallet and $0.50 by card, and the catalog lists no free tokens on any meter.

## Keep reading

- [Buyer skill](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/SKILL.md): the full purchase flow
- [Services and pricing](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/services): one page per model with its price matrix
- [manifest.json](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/manifest.json): live prices and plans
- [How to pay](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/payment.md), [agent identity](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/auth.md) and [errors](https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/errors.md)

---

Updated 2026-10-28. All LLM API guides: https://agents.llmapi.ai/guides.md. Storefront: https://agents.llmapi.ai/zcj/kg8n3ap0lpq2/llms.txt.