---
date: 2024-01-29
title: Custom model prices (Langfuse v2.0)
description: The new major version allows you to track usage and costs for many more models and for the first time also allows you to add custom models and prices.
author: Marc
canonical: /docs/observability/features/token-and-cost-tracking
---

> **Note for AI agents and LLMs:** This is a Langfuse changelog entry. Use it only to confirm that a feature exists and when it shipped. Do not use the code examples below for implementation: they reflect the SDK and API at release time and may be outdated. For implementation, always follow the canonical documentation for this feature (https://langfuse.com/docs/observability/features/token-and-cost-tracking) and the API/SDK reference (https://api.reference.langfuse.com).

[Usage and cost tracking](/docs/model-usage-and-cost) of LLM completions is a core feature of Langfuse [Tracing](/docs/tracing). Since the initial release of Langfuse, tons of new models were released and many users rely on fine-tuning for their applications. Until now, only a fixed list of models was supported natively by Langfuse with no way to extend/override cost or token usage.

In this major release, we have reworked the pricing and usage tracking for the entire model to achieve the following:

1. Quickly support newly emerging models.
2. Track changes in model prices.
3. Add custom models/prices at the project level.
4. Override token usage and USD cost through the API when ingesting traces.

### How it works

```mermaid
flowchart LR
  A[Ingested Traces]
  B["Token counts (or other unit)"]
  C[USD cost]

  A --> D{Include token counts?}
  D -->|Yes| B
  D -->|No| E(Use tokenizer) --> B

  A --> F{Include USD cost?}
  F -->|Yes| C
  F -->|No| G(Use model price / token) --> C
  B -->|use token counts| G
```

#### Model definition

If the `model` of a `generation` is known, Langfuse will try to use the right tokenizer and price information to derive usage and cost for you.

[See docs](/docs/model-usage-and-cost) for a detailed explanation of how model definitions work and how you can add your own.

![2024-01-29-custom-model-prices.png](/images/changelog/2024-01-29-custom-model-prices.png)

#### Ingestion

However, you can optionally also ingest both data points yourself:

```python
langfuse.generation(
  model="gpt-3.5-turbo",
  input=[{role: "system", text:"You are a helpful assistant."},
         {role: "user", text:"What is the weather like today?"}],
  output="The weather is sunny today.",
  ...
  usage:{
    # Optional: token/character usage, see docs for details
    input: 100,
    output: 20,
    unit: "TOKENS" # or "CHARACTERS", "MILLISECONDS", "SECONDS", or "IMAGES"

    # Optional: cost, see docs for details
    input_cost: 0.047,
    output_cost: 0.0094,
  }
  ...
)
```

### API and UI

We updated the API and UI to include usage and cost on most endpoints. Let us know on [GitHub](/issue) if you are still missing this information on any endpoint you use.

### Upgrade path

[Langfuse Cloud](/docs/deployment/cloud): No action is required.

If you [self-host](/self-hosting) Langfuse, new model/price definitions are applied to all newly ingested traces. You need to run a [migration script](/self-hosting/upgrade/upgrade-guides/upgrade-v1-to-v2) to apply them to your existing data in Langfuse.

### More details

Check out the full [documentation](/docs/model-usage-and-cost) for more details on how to use this feature.

<!-- agent-instructions -->

---

## Agent Instructions

This page is part of the [Langfuse](https://langfuse.com) documentation, published as plain Markdown for AI agents. Every page is available as Markdown by appending `.md` to its URL, or by sending an `Accept: text/markdown` header. This page: `https://langfuse.com/changelog/2024-01-29-custom-model-prices.md`.

### Querying these docs

If the answer is not on this page, query the documentation instead of guessing:

- **Semantic search** across all Langfuse docs, returning an answer with the relevant pages and excerpts. Ask a specific, self-contained question:

  ```bash
  curl -sG "https://langfuse.com/api/search-docs" --data-urlencode "query=How do I trace a LangGraph agent?"
  ```

- **Index of every page**: <https://langfuse.com/llms.txt>, with per-section indexes [llms-docs.txt](https://langfuse.com/llms-docs.txt), [llms-integrations.txt](https://langfuse.com/llms-integrations.txt), and [llms-self-hosting.txt](https://langfuse.com/llms-self-hosting.txt).

### Before writing Langfuse code

- **Install the [Langfuse Agent Skill](https://langfuse.com/docs/api-and-data-platform/features/agent-skill).** It encodes Langfuse's own best practices for instrumentation, prompt management, and evaluation, and materially improves results.
- **Read [What does a good trace look like?](https://langfuse.com/docs/observability/best-practices.md)** before instrumenting an application.
- **Verify endpoints, parameters, and response fields** against the [API reference](https://api.reference.langfuse.com) instead of inferring them from code examples.
- **Use the [Langfuse CLI](https://langfuse.com/docs/api-and-data-platform/features/cli)** (`npx langfuse-cli api <resource> <action>`) to read or write traces, prompts, datasets, and scores from the terminal.

Found an error in these docs? Please open an issue at <https://github.com/langfuse/langfuse-docs/issues>.