Breaking

Meta Launches Its Own Image Model Muse Image on Meta AI

July 7, 2026 at 15:11 EDT

Media Generation
Foundation Models

On July 7, 2026, Meta officially rolled out "Muse Image," the first dedicated image-generation model from Meta Superintelligence Labs (MSL), making it available across Meta AI, Instagram and WhatsApp. The model is designed to understand complex prompts through reasoning and to seamlessly blend multiple photos.

July 7, 2026 · Meta Superintelligence Labs

Meta unveils Muse Image, its first in-house image-generation model

Announced by Chief AI Officer Alexandr Wang and live inside Meta AI, Muse Image blends multiple photos and uses advanced reasoning to interpret complex prompts — moving Meta from external providers toward proprietary, agentic generation.

How it works together

Muse Spark

Multimodal reasoning: tool use, web search, planning & multi-agent orchestration

→

Muse Image

Understands intent, organizes steps & renders high-quality images

→

Agentic output

Beyond plain text-to-image — download, share, presets & sketch editing

Companion model — Muse Spark benchmarks

Standalone benchmarks for Muse Image were not published at announcement.

58%

Humanity's Last Exam
(Contemplating mode)

42.8%

HealthBench Hard

38%

FrontierScience Research

262k

token context window

Muse Spark accuracy, to scale

Score on each reasoning benchmark (%)

58%

Humanity's
Last Exam

42.8%

HealthBench
Hard

38%

FrontierScience
Research

Enthusiasts say

Agentic reasoning yields high first-pass accuracy
Strong on complex scenes — interior design, exploded diagrams, photorealistic scenes from web references

Skeptics caution

Head-to-head vs Flux, Midjourney & Ideogram still to come
Samples promising, but assessment remains limited — no standalone benchmarks or API preview yet

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.