> ## Documentation Index
> Fetch the complete documentation index at: https://docs.4096bytes.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Supported models and pricing

> Review the current model calling costs shown by 4096bytes

This page shows input, output, cache, official price, and discount multiplier values in USD per `1M tokens`, so you can estimate Codex, Claude Code, or SDK usage cost.

<Warning>
  Model availability, prices, discounts, and labels may change by plan, group, or release date. Before making requests, use the current values shown in the dashboard.
</Warning>

## Claude Code - Standard

Only available in Claude Code, with the Claude Code Pro + Max pool. The dashboard currently shows `9` models, with a `0.3x` discount multiplier.

| Model                      | Input price (1M tokens) | Output price (1M tokens) | Cache price (1M tokens)                | Official price (1M tokens) | Discount multiplier |
| -------------------------- | ----------------------- | ------------------------ | -------------------------------------- | -------------------------- | ------------------- |
| claude-opus-4-8            | \$1.5                   | \$7.5                    | Cache write: $1.875; Cache read: $0.15 | Input: $5; Output: $25     | 0.3x                |
| claude-opus-4-7            | \$1.5                   | \$7.5                    | Cache write: $1.875; Cache read: $0.15 | Input: $5; Output: $25     | 0.3x                |
| claude-opus-4-6            | \$1.5                   | \$7.5                    | Cache write: $1.875; Cache read: $0.15 | Input: $5; Output: $25     | 0.3x                |
| claude-opus-4-5-20251101   | \$1.5                   | \$7.5                    | Cache write: $1.875; Cache read: $0.15 | Input: $5; Output: $25     | 0.3x                |
| claude-sonnet-4-6          | \$0.9                   | \$4.5                    | Cache write: $1.125; Cache read: $0.09 | Input: $3; Output: $15     | 0.3x                |
| claude-sonnet-4-5-20250929 | \$0.9                   | \$4.5                    | Cache write: $1.125; Cache read: $0.09 | Input: $3; Output: $15     | 0.3x                |
| claude-sonnet-4-20250514   | \$0.9                   | \$4.5                    | Cache write: $1.125; Cache read: $0.09 | Input: $3; Output: $15     | 0.3x                |
| claude-haiku-4-5-20251001  | \$0.3                   | \$1.5                    | Cache write: $0.375; Cache read: $0.03 | Input: $1; Output: $5      | 0.3x                |
| claude-haiku-4-5           | \$0.3                   | \$1.5                    | Cache write: $0.375; Cache read: $0.03 | Input: $1; Output: $5      | 0.3x                |

## GPT Codex - Standard

Supports the Codex client and third-party clients. It also supports image generation with `image-2`. The dashboard currently shows `10` models, with a `0.1x` discount multiplier.

| Model                       | Input price (1M tokens) | Output price (1M tokens) | Cache price (1M tokens) | Official price (1M tokens) | Discount multiplier |
| --------------------------- | ----------------------- | ------------------------ | ----------------------- | -------------------------- | ------------------- |
| gpt-image-2                 | \$0.5                   | \$1                      | Cache read: \$0.125     | Input: $5; Output: $10     | 0.1x                |
| gpt-5.5-openai-compact      | \$0.5                   | \$3                      | Cache read: \$0.05      | Input: $5; Output: $30     | 0.1x                |
| gpt-5.5                     | \$0.5                   | \$3                      | Cache read: \$0.05      | Input: $5; Output: $30     | 0.1x                |
| gpt-5.4-openai-compact      | \$0.25                  | \$1.5                    | Cache read: \$0.025     | Input: $2.5; Output: $15   | 0.1x                |
| gpt-5.4-mini-openai-compact | \$0.25                  | \$1.5                    | Cache read: \$0.025     | Input: $2.5; Output: $15   | 0.1x                |
| gpt-5.4-mini                | \$0.075                 | \$0.45                   | Cache read: \$0.0075    | Input: $0.75; Output: $4.5 | 0.1x                |
| gpt-5.4                     | \$0.25                  | \$1.5                    | Cache read: \$0.025     | Input: $2.5; Output: $15   | 0.1x                |
| gpt-5.3-codex-spark         | \$0.175                 | \$1.4                    | Cache read: \$0.0175    | Input: $1.75; Output: $14  | 0.1x                |
| gpt-5.3-codex               | \$0.175                 | \$1.4                    | Cache read: \$0.0175    | Input: $1.75; Output: $14  | 0.1x                |
| codex-auto-review           | \$0.5                   | \$3                      | Cache read: \$0.05      | Input: $5; Output: $30     | 0.1x                |

## Field reference

| Field                      | Meaning                                                                           |
| -------------------------- | --------------------------------------------------------------------------------- |
| Input price (1M tokens)    | Cost for input processed by the model.                                            |
| Output price (1M tokens)   | Cost for generated output.                                                        |
| Cache price (1M tokens)    | Cost for cache writes or cache reads. Some models only show cache read pricing.   |
| Official price (1M tokens) | Official input and output list prices, used to compare against discounted prices. |
| Discount multiplier        | Current group multiplier compared with the official price.                        |

## Usage tips

* Estimate input, output, cache write, and cache read volume separately.
* Choose a model by price, task difficulty, response quality, and client support.
* When you configure a client or SDK, copy the full model name from the dashboard.
* If the API Key group does not include the target model, adjust the group in the dashboard or create a new API Key.

## Next step

Continue to [Get an API Key](/en/quickstart/get-api-key), or choose the [client](/en/clients/codex) you want to connect.