Latest news and updates from LLM Gateway
Decision models return probabilities instead of prose, but trying one meant hand-writing JSON. Lounge's new Decision Studio lets you paste a ticket, build yes/no, pick-one and rating questions in the browser, compare GPT-6 Luna Decisions, Microsoft-Decision-1 and Jev on the same input, and copy the run as a /v1/decisions request.
DevPass Code 2.0 adds the Smart Router: before each message, an LLM Gateway decision model reads your request and sends it to Claude Haiku, Sonnet or Opus. You choose the decision model, GPT-6 Luna Decisions, Microsoft-Decision-1 or Jev, and session tabs let you run several agents side by side.
LLM Gateway now serves the OpenAI Decisions API at /v1/decisions. GPT-6 Luna Decisions answers typed questions about text and images at $0.10 per million input tokens, Microsoft-Decision-1 on Azure and Jev run on the same request shape, and the OpenAI SDKs work with a base URL change.
Every new AI model released between September 28 and October 11, 2026 and already live on LLM Gateway: Claude Haiku 5.5 and Sonnet 5.5, GPT-6.1 Sol, Mistral Large 4, GLM 5.3 Fast, Reka Edge 2603, Gemini Nano Banana 2.1, Hy Image 3.5 Preview, Ling 3.1 Flash and Grok Imagine Video 1.5 Lite, with prices, context windows and the model ID to call.
Executive Order 14434 tells federal agencies to say Super Intelligence (SI) instead of Artificial Intelligence (AI). Here is what the order actually changes, what it does not, and how to run every SI model through one SI gateway.
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.
By Ismail Ghallou
September 30, 2026
Use the Perplexity Search API through LLM Gateway to get ranked web results with extracted page content for agents and retrieval pipelines, billed per search with the same key you use for every model.
Automatic model selection now reads the request before it routes it. Choose the models `smart` may resolve to, and a classifier rates each prompt so a one-line fix goes to a cheap model and a hard problem goes to a frontier one. Available to every organization while in beta.
Compliance alerts watch the models your provider compliance policy blocks today and notify your team the moment a provider that meets your requirements starts serving one — in-app, by email, or in Slack. Downgrade alerts fire when a provider stops meeting the policy. Available on the Enterprise plan.
The SCX model discount was due to end on September 24. It now runs 15 more days: 20% off the same seven models on LLM Gateway until October 9, 2026. Nothing changes in how you route requests.
Most classification and routing calls use a chat model, then parse prose and hope the JSON holds. The new /v1/systemone LLM decision API on LLM Gateway returns typed yes/no, choice, and score answers with calibrated probabilities, billed on input tokens only.
Perplexity retires its Sonar chat completions API on September 27, 2026. On LLM Gateway, perplexity/sonar moves to Perplexity's Agent API on September 25 with no changes to your requests and lower prices, while sonar-pro and sonar-reasoning-pro stop being routable on September 27. Here is exactly what changes, why, and what to do about it.
The SCX model discount brings 20% off seven selected models on LLM Gateway, starting when the Runware promotion ends on September 9 and now extended through October 9. See the eligible models and how to route requests through SCX.
Airside is the self-serve carrier console where LLM providers claim their listing on LLM Gateway, register models, file prices for review, and tune the discount and margin that win routed traffic. Listing costs a one-time $2,500 fee per provider company, and every model is verified live before it goes on the departure board, with a per-request breakdown of what your endpoint served.
Switching off the Vercel AI Gateway used to mean rewriting how your AI SDK app resolves models — and losing provider-native web search on the way out. LLM Gateway now implements the AI SDK's own gateway protocol, so the migration is one line and your bare model strings keep working.
LLM Gateway Chat now has Projects: group related chats, upload files — PDFs and spreadsheets included — as a knowledge base, and get answers grounded in your own documents via RAG, with source citations, on any of 280+ models. Projects also remember durable facts across chats. Available to every Chat user.
DevPass Code is an open-source terminal coding agent that talks only to LLM Gateway. One browser login, every model, and no per-provider API keys to juggle. Use it pay-as-you-go or on a DevPass coding subscription.
Three months of updates: speech generation and Audio Studio, OpenAI-compatible embeddings, OCR, DevPass coding plans, chat subscription plans, enterprise IAM and master keys, SOC 2 Type II, 40+ new models, and much more.
Most dashboards show what you spent, not where it went. LLM Gateway's enterprise LLM analytics break cost, requests, and tokens down by model, project, API key, and team member — no data warehouse to build. Member and organization-wide analytics are available on the Enterprise plan.
LLM Gateway has successfully completed its SOC 2 Type II audit. Here's what that means, why it matters for teams routing LLM traffic through us, and how to request our report.
Nine text-to-speech models from ElevenLabs, OpenAI, and Google are now one OpenAI-compatible API call away — plus a new Audio Studio in the Playground to compare voices and models side by side.
LLM Gateway offers exclusive discounts on DeepSeek V3.2, V3.1, and R1 through partner providers — up to 30% off base pricing, applied automatically to every request.
Three months of updates: video generation, Image Studio, sessions, GPT-5.4 family, enterprise guardrails, 5+ new providers, and much more.
LLM Gateway partners with Alibaba Cloud to bring you 20% off 26 Qwen AI models — including Qwen3 Max, Qwen3 Coder, QwQ reasoning, vision-language, and image generation models.
Three months of updates: 15+ new models, team management, referral program, tiered pricing, data retention, new providers, and much more.
Use the CYBERMONDAY promo code to get 50% off credits for a limited time.
Use the BLACKFRIDAY promo code to get 50% off credits for a limited time.
Compare GPT-5, Claude, Gemini, and 280+ other models side by side. Generate images, test prompts, and find the best model for your use case.
Connect your internal LLM deployments or any OpenAI-compatible API to LLM Gateway—and get the same analytics, caching, and routing.
One API for 280+ models across 35+ providers. Route requests, track costs, and switch models without changing your code.