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

# LLM Models & Pricing

> Complete pricing and model limits reference for all supported LLM providers

## Anthropic models

**Pricing (per 1M tokens):**

| Model                       | Input  | Output  | Cache Write | Cache Read |
| --------------------------- | ------ | ------- | ----------- | ---------- |
| `claude-opus-4-6`           | \$5.00 | \$25.00 | \$6.25      | \$0.50     |
| `claude-sonnet-4-6`         | \$3.00 | \$15.00 | \$3.75      | \$0.30     |
| `claude-haiku-4-5-20251001` | \$1.00 | \$5.00  | \$1.25      | \$0.10     |

**Model limits:**

| Model                       | Max Prompt (chars) | Max Chunk Functions | Max Output Tokens |
| --------------------------- | ------------------ | ------------------- | ----------------- |
| `claude-opus-4-6`           | 400,000            | 120                 | 16,384            |
| `claude-sonnet-4-6`         | 400,000            | 120                 | 16,384            |
| `claude-haiku-4-5-20251001` | 400,000            | 120                 | 16,384            |

Anthropic models have a 200k token context window (\~800k chars). Kong uses prompt caching to reduce costs on repeated system prompt content.

## OpenAI models

**Pricing (per 1M tokens):**

| Model         | Input   | Output  | Cached Input |
| ------------- | ------- | ------- | ------------ |
| `gpt-4o`      | \$2.50  | \$10.00 | \$1.25       |
| `gpt-4o-mini` | \$0.15  | \$0.60  | \$0.075      |
| `o1`          | \$15.00 | \$60.00 | \$7.50       |
| `o3-mini`     | \$1.10  | \$4.40  | \$0.55       |

**Model limits:**

| Model         | Max Prompt (chars) | Max Chunk Functions | Max Output Tokens |
| ------------- | ------------------ | ------------------- | ----------------- |
| `gpt-4o`      | 350,000            | 80                  | 16,384            |
| `gpt-4o-mini` | 350,000            | 80                  | 16,384            |
| `o1`          | 400,000            | 40                  | 32,768            |
| `o3-mini`     | 400,000            | 60                  | 32,768            |

OpenAI standard models have 128k token context (\~512k chars). Reasoning models (`o1`, `o3-mini`) have 200k context but use smaller batch sizes due to higher per-call cost.

## Custom providers

No pricing is tracked for custom endpoints. Token counts are still recorded and displayed in the analysis summary. See [Custom Endpoints](/configuration/custom-endpoints) for configuration.

## Cost estimation

Rough cost ranges by binary size (using Claude Opus as baseline):

| Binary size        | Typical cost | Typical time |
| ------------------ | ------------ | ------------ |
| \~300 functions    | \$10-50      | 5-15 min     |
| \~1,000 functions  | \$40-150     | 20-60 min    |
| \~3,000+ functions | \$150-500+   | 1-3 hours    |

These vary significantly with:

* **Model choice** — `gpt-4o-mini` is \~30x cheaper than `claude-opus-4-6`
* **Binary complexity** — obfuscated code requires more tool calls
* **Signature matches** — known library functions skip LLM analysis entirely

## Further reading

* [LLM Providers](/configuration/llm-providers) — provider setup and configuration
* [Custom Endpoints](/configuration/custom-endpoints) — using local models