AI models

Every major model.
One budget per agent.

The AI models of the LiveContext catalog, and when each one landed. Bring your own key and pay the provider directly, or run on ours and pay in credits. Swap the model on any agent, workflow step or chat without touching anything else you built.

The catalog holds 360 models across 12 providers, plus 458 more reachable through OpenRouter. Listed below are the 102 of them: one row per model family rather than one per moving alias or dated snapshot, and only where the provider published a release date.

8models on this page
1Mwidest context window
$0.25cheapest input, per 1M
2025oldest release listed
Nov '25
Feb '26
Mar '26
May '26
Jul '26
Aug '26
Sep '26
ModelProviderReleasedContextInput / 1MOutput / 1MCan
Gemini 3.8 Flashgemini-3.8-flash · Google's newest workhorse Flash.GoogleSep 2, 20261M$0.75$3.75ReasoningVisionToolsCacheWeb
Gemini 3.7 Flashgemini-3.7-flashGoogleAug 13, 20261M$0.75$3.75ReasoningVisionToolsCacheWeb
Gemini 3.6 Flashgemini-3.6-flashGoogleJul 21, 20261M$0.75$3.75ReasoningVisionToolsCacheWeb
Gemini 3.5 Flash-Litegemini-3.5-flash-lite · Lowest-latency Gemini.GoogleJul 21, 20261M$0.30$2.50ReasoningVisionToolsCacheWeb
Gemini 3.5 Flashgemini-3.5-flash · Shipped at Google I/O 2026.GoogleMay 19, 20261M$1.50$9ReasoningVisionToolsCacheWeb
Gemini 3.1 Flash-Litegemini-3.1-flash-liteGoogleMar 20261M$0.25$1.50ReasoningVisionToolsCacheWeb
Gemini 3.1 Progemini-3.1-pro-preview · Newest Pro-tier Gemini in the API.GoogleFeb 19, 20261M$2$12ReasoningVisionToolsCacheWeb
Gemini 3 Progemini-3-pro-preview · Opened the Gemini 3 generation.GoogleNov 18, 20251M$2$12ReasoningVisionToolsCacheWeb
Extended reasoning / thinkingReads imagesCalls tools and integrationsPrompt caching, so repeat context costs lessBuilt-in web search

8 of 102 models listed here. One row per model family: moving aliases, dated snapshots of a model already listed, and rows whose provider published no release date are left out. Context windows, list prices and capabilities come from the live catalog the app itself reads; release dates come from the providers' own announcements.

A model shows up here once it is in the catalog, so this page follows the platform rather than a marketing calendar. How keys, CLI bridge providers and per-model settings work is covered in the models documentation.