# How Decagon uses Modal

**65%**: “The result was a 65% reduction in total response time”

As published on [decagon.ai](https://decagon.ai/blog/real-time-voice-ai-on-modal), November 2025. Captured by usedby on 2026-11-17.

- Company: [Decagon](https://www.usedby.ai/companies/decagon.md)
- Tool: [Modal](https://www.usedby.ai/tools/modal.md)
- Industry: [Artificial Intelligence](https://www.usedby.ai/companies/industry/artificial-intelligence.md)

## The use case: how Decagon deploys Modal

Decagon uses Modal's infrastructure to train compact open-source models with fine-tuning and reinforcement learning, and to serve them for real-time voice AI. Modal's inference team also built a custom draft model and an asynchronous scheduler for speculative decoding in SGLang to reduce latency.

Summary written by usedby from the source page, in English. The figures are those of Modal and Decagon, not ours.

> Modal's infrastructure powered this progress, enabling Decagon to train and serve increasingly capable models while scaling seamlessly to meet rapidly growing traffic.

- **38%**: “The resulting draft model achieved 38% higher accept lengths than open-source baselines, unlocking a major leap in inference efficiency.”
- **12%**: “These changes completely eliminated idle time on H200 GPUs and improved throughput by up to 12%.”

## Other AI tools used by Decagon

- [Together AI](https://www.usedby.ai/tools/together-ai.md) — Confirmed
- [Cartesia](https://www.usedby.ai/tools/cartesia.md) — Observed
- [Centralize](https://www.usedby.ai/tools/centralize.md) — Observed
- [Claude](https://www.usedby.ai/tools/claude.md) — Observed
- [Cursor](https://www.usedby.ai/tools/cursor.md) — Observed
- [Deepgram](https://www.usedby.ai/tools/deepgram.md) — Observed

## Modal users in Artificial Intelligence

- [Glean](https://www.usedby.ai/companies/glean.md) — Declared by the company
- [Hebbia](https://www.usedby.ai/companies/hebbia.md) — Declared by the company
- [Runway](https://www.usedby.ai/companies/runway.md) — Declared by the company
- [Achira](https://www.usedby.ai/companies/achira.md) — Observed
- [Applied Compute](https://www.usedby.ai/companies/applied-compute.md) — Observed
- [Cartesia](https://www.usedby.ai/companies/cartesia.md) — Observed

## What usedby checked

We compared the story with its live page on 2026-11-17.

- Checked: the figure 65% is printed word for word on the page, near the name of Decagon.
- Checked: the passage quoted above is copied word for word from the page, near the name of Decagon.
- Checked: the publication date is read from the page’s own metadata, never guessed.
- Checked: Decagon uses Modal. Confirmed line. Latest check across sources: 2026-11-17.
- Not checked: the result itself. We quote it; we did not measure it.

---
Source: https://www.usedby.ai/case-studies/decagon-modal · How we check: https://www.usedby.ai/methodology
