---
title: "Q3 2026: Smarter Routing, Airside & Realtime"
subtitle: "LLM Gateway product roundup — 50M requests and 1.805T tokens"
slug: q3-2026-smarter-routing-airside-and-realtime
url: https://listedarticles.com/articles/q3-2026-smarter-routing-airside-and-realtime
canonical_url: https://llmgateway.io/blog/q3-2026-roundup
content_type: changelog
language: en
published_at: 2026-09-30T00:00:00.000Z
updated_at: 2026-10-02T00:12:57.254Z
author: "Ismail Ghallou"
author_url: https://llmgateway.io/
authored_by: human
publisher: "LLM Gateway"
publisher_url: https://llmgateway.io/
topics: ["AI", "LLMs", "Infrastructure", "Developer Tools", "Announcements"]
license: all-rights-reserved
word_count: 285
reading_minutes: 1
citation: "Ismail Ghallou, LLM Gateway. \"Q3 2026: Smarter Routing, Airside & Realtime.\" 30 Sept 2026. https://llmgateway.io/blog/q3-2026-roundup (all-rights-reserved)"
# The full text follows. The web page shows an extract and sends readers
# to the source above; quote the citation and link the canonical URL.
---

# Q3 2026: Smarter Routing, Airside & Realtime

*LLM Gateway product roundup — 50M requests and 1.805T tokens*

> LLM Gateway's Q3 2026 roundup: smart and cache-aware routing, Airside provider console, realtime voice, Lounge projects, coding-tool integrations, and stronger enterprise controls—with a public traffic snapshot.

# Q3 2026: Smarter Routing, Airside & Realtime

The Q3 2026 product roundup: smart and cache-aware routing, Airside, realtime voice and transcription, Lounge projects, coding tools, and stronger enterprise controls. Plus a public traffic snapshot: 50 million requests and 1.805 trillion tokens.

## Q3 2026 by the numbers

Across the platform (July 1–September 30, 2026 UTC snapshot):

| Metric | Value |
| --- | --- |
| Requests recorded | 50,036,094 |
| Total tokens | 1.805 trillion |
| Busiest day | 1,576,727 requests on September 25 |

Top models by requests included gemini-embedding-2, deepseek-v4-flash, and gpt-5.4-mini. Top providers by requests: OpenAI, Google Vertex, Google AI Studio, Azure, DeepSeek.

## Let the workload choose the model

Smart routing adds `model: "smart"`: choose eligible models, then pick cheapest or use a classifier for difficulty/task/output type. Sessions can adapt between turns; decisions are visible in request details. Provider selection also became adaptive and cache-aware. Dynamic routes (Enterprise) add named flows with conditionals, weighted splits, and rollback.

## Airside

Airside is a self-service console for model providers: claim a carrier, register models, preflight-verify capabilities, submit changes for review, and manage listings without spreadsheet chains.

## Realtime, Lounge, and coding tools

- Realtime voice via OpenAI-compatible WebSocket at `wss://api.llmgateway.io/v1/realtime`, with ephemeral client secrets and transcription-only sessions.
- Lounge (formerly chat) adds projects with knowledge bases, memory, and citations.
- DevPass Code, `llmgateway launch`, browser login, organization skills, and an official VS Code extension for Copilot Chat.

## More API surfaces

Perplexity Search at `/v1/search`, System One decisions at `/v1/systemone`, AI SDK gateway-protocol compatibility, image quality controls, and video generation via `@llmgateway/ai-sdk-provider` v4.

## Enterprise controls

Zero data retention, provider headquarters restrictions, compliance alerts, project-scoped developers and teams, per-member usage limits, project-level guardrail overrides, and hash-only gateway key storage.

*Original: [llmgateway.io/blog/q3-2026-roundup](https://llmgateway.io/blog/q3-2026-roundup)*