OpenAI's GPT-5.6 Sol Launches Thursday, Plus Meta's Muse, Claude Cowork & Higgsfield Apps

OpenAI's GPT-5.6 Sol Launches Thursday, Plus Meta's Muse, Claude Cowork & Higgsfield Apps

0:00 / 3:20

Chapters

DAILY ROUNDUP

OpenAI's GPT-5.6 Sol Launches Thursday, Plus Meta's Muse, Claude Cowork & Higgsfield Apps

calendar_today Date:
schedule Duration: 3:20

OpenAI's three-tier GPT-5.6 launches Thursday, Meta ships its first image model Muse, Claude Cowork goes mobile, and Higgsfield turns one image into a working app.

  • 01. OpenAI's GPT-5.6 Sol, Terra and Luna launch publicly Thursday, undercutting the outgoing generation on price.
  • 02. Meta Superintelligence Labs released Muse Image, its first in-house media generation model, with Muse Video coming soon.
  • 03. Claude Cowork now follows tasks from desktop to mobile and web, with beta starting on the Max plan.
  • 04. Higgsfield Apps lets Claude Fable 5 engineer a working app from a single reference image.
OpenAI is taking GPT-5.6 public this Thursday as a three-tier family rather than a single model. Sol is the flagship, Terra is a cheaper mid-tier option pitched as competitive with GPT-5.5 at half the cost, and Luna is the fast, low-cost model built for high-volume work. Pricing ranges from five dollars per million input tokens for Sol down to one dollar for Luna, undercutting the previous generation across the board. Sol also comes with a tightened safety stack, including stricter checks on higher-risk requests and repeated misuse. Preview access is expanding globally ahead of Thursday's full public rollout. Meta Superintelligence Labs has released Muse Image, its first in-house media generation model. It follows instructions closely, edits with precision, composes from multiple reference images, and can call its own search and coding tools to refine results before output. It draws on Instagram for social context and integrates with Muse Spark for agentic media workflows. Muse Image is live now in the Meta AI app, on Instagram Stories in the US, and on WhatsApp in limited countries. A matching video model, Muse Video, is coming soon. This marks Meta's first serious in-house entry into the image generation race, an area where it has previously relied on other providers. Anthropic is expanding Claude Cowork beyond desktop to mobile and the web, so tasks handed off at a desk keep running after the laptop closes. Users can check progress from their phone, and when Claude reaches a decision that needs human input, that question now follows the user to mobile as well. Anthropic's own usage data shows over 90% of Cowork activity isn't software development at all, but on-the-fly business operations requests. The beta starts with Max plan subscribers over the coming weeks, with other plans to follow later. Higgsfield has launched Higgsfield Apps, which generates a working app built around any of its image or video models from a single reference image. Claude Fable 5 handles the engineering side, covering architecture and front-end polish, while Higgsfield supplies the visual assets and hosts the result. Output can be shipped as a full website, or packaged as a browser extension, plugin, or desktop app. Higgsfield is calling it their biggest release yet, though as with any self-declared milestone, the real test will be what developers actually build with it.
Stories In This Briefing

3 stories covered

VIEW ALL NEWS
Related Stories

More on these topics