v0.25.3

mahimailabs/voicegatewayv0.25.3Aug 19, 2026by mahimairaja

AI Summary

This release introduces a new read endpoint for turn percentiles to ensure consistency between baseline modes and refactors pricing logic to evaluate models before execution.

Key Highlights

New turn-percentile read endpoint for baseline mode consistency
Pricing logic updated to evaluate models before execution
LLM pricing now represented as two distinct numbers

New Features

Turn-percentile read endpoint
Pre-run model pricing
Dual-number LLM pricing representation

Full Release Notes

## Docker Images

Published to Docker Hub:

- `mahimairaja/voicegateway:0.25.3`
- `mahimairaja/voicegateway:latest`
- `mahimairaja/voicegateway-dashboard:0.25.3`
- `mahimairaja/voicegateway-dashboard:latest`

Built for `linux/amd64` and `linux/arm64`.

## Quick Start

```bash
docker run -p 8080:8080 \
  -v $(pwd)/voicegw-data:/data \
  mahimairaja/voicegateway:0.25.3
```

See [docs.voicegateway.dev](https://docs.voicegateway.dev) for full documentation.


## What's Changed
* Add a turn-percentile read endpoint so both baseline modes agree by @mahimairaja in https://github.com/mahimailabs/voicegateway/pull/259
* Price a model before it runs, and price an LLM as two numbers by @mahimairaja in https://github.com/mahimailabs/voicegateway/pull/260


**Full Changelog**: https://github.com/mahimailabs/voicegateway/compare/v0.25.2...v0.25.3