Model Updates

New and updated AI models from every major provider.

This Week

Perplexitymodel

Link·Jul 10, 2026

GPT-5.6 Sol and Terra in Perplexity

The two stronger GPT-5.6 tiers from OpenAI are now selectable in Perplexity and in Perplexity Computer.

Best for: Putting GPT-5.6 to work against live web sources without leaving Perplexity.
When to use it: You want OpenAI reasoning applied to current information rather than to model memory.
Who benefits: Perplexity users who compare model outputs on the same question.

OpenAImodel

Link·Jul 9, 2026

GPT-5.6 in the API (Generally Available)

The GPT-5.6 family from OpenAI is out of limited preview and open to all API developers in three tiers.

Best for: Matching a price and speed tier to each task instead of routing everything through one flagship model.
When to use it: You are picking a model for production and want cost to track how hard the task actually is.
Who benefits: Developers and technical marketers building on the OpenAI API.

Perplexitymodel

Link·Jul 9, 2026

Perplexity Computer Orchestrator Model (Research Preview)

A cheaper model that runs the planning layer inside Perplexity Computer, released as a research preview.

Best for: Running agent tasks in Perplexity Computer without paying frontier prices on every step.
When to use it: Cost per task matters more to you than squeezing out the last increment of quality.
Who benefits: Perplexity Computer users running long or repetitive agent jobs.

Metamodel

Link·Jul 9, 2026

Muse Spark 1.1

A multimodal reasoning model from Meta built for agent work, now in the Meta AI app and open to developers through a new API in public preview.

Best for: Agent work that has to read screens, documents, and video, then act across applications.
When to use it: The task needs a model to operate software and hold a long project in context.
Who benefits: Developers building agents, and Meta AI users who switch on Thinking mode.

OpenAImodel

Link·Jul 8, 2026

GPT-Live

A new generation of ChatGPT voice models that can listen and talk at the same time, so a conversation feels closer to talking with a person.

Best for: back-and-forth voice conversations where you interrupt, pause, and think out loud instead of typing
When to use it: brainstorming by talking, live translation, or asking questions while your hands are busy
Who benefits: marketers who work through ideas and research out loud rather than at a keyboard

Metamodel

Link·Jul 7, 2026

Muse Image and Muse Video

Meta's new image and video generation models, free inside the Meta AI app, Instagram, and WhatsApp.

Best for: Generating and editing marketing visuals and short video from a text prompt or reference images, without a separate design tool
When to use it: You need an on-brand image or a short clip fast and already work inside Meta's apps
Who benefits: Social and content marketers creating for Instagram, Facebook, and WhatsApp

OpenAImodel

Link·Jul 6, 2026

GPT-Realtime-2.1-mini

A cheaper voice model in OpenAI's API that now handles reasoning and tool use, for developers building live voice apps.

Best for: Building real-time voice features like agents and assistants where cost matters but you still want the model to reason and call tools
When to use it: You need a voice model that can think through steps without paying for the full-size Realtime model
Who benefits: Developers building live voice products on OpenAI's Realtime API

Last Week

Microsoft Copilotmodel

Link·Jul 1, 2026

MAI-Image-2-Efficient in Copilot for PowerPoint

A faster Microsoft image model for generating slide visuals inside Copilot in PowerPoint.

Best for: generating and iterating on images for slides without leaving PowerPoint
When to use it: you are drafting a deck and want visuals in the moment rather than sourcing them separately
Who benefits: anyone building presentations in Microsoft 365 with Copilot

Perplexitymodel

Link·Jun 30, 2026

Claude Sonnet 5 in Perplexity

Perplexity added Anthropic's Claude Sonnet 5 as a selectable model for paid subscribers.

Best for: running Perplexity searches and tasks on Anthropic's newest Sonnet model instead of the default
When to use it: you prefer Claude's reasoning and writing for research inside Perplexity
Who benefits: Perplexity Pro and Max subscribers who want to pick the underlying model

Googlemodel

Link·Jun 30, 2026

Gemini Omni Flash (Public Preview)

A cost-efficient version of Google's Omni model for generating video and editing it through conversation.

Best for: Generating short video and refining it by describing changes in plain language, at lower cost
When to use it: You want video generation or conversational editing in a workflow without the base Omni model's cost
Who benefits: Developers and creative teams building video features on the Gemini API