Meta releases Muse Spark 1.3 agentic coding model
Meta Superintelligence Labs' fourth Muse Spark release in five months uses ~20% fewer tool calls and ~25% fewer tokens than Muse Spark 1.2. It scores 62 on Artificial Analysis vs 59 for Gemini 3.8 Flash, with pricing at $0.2 vs $25 per 1M output tokens versus Claude Opus 5.
People · Mark Zuckerberg
How this story unfolded
2 days · 6 reports · 25 community posts · from Sep 2
- Sep 2
- Sep 3
[AINews] Muse Spark 1.3 matches GPT-5.6-Sol, confirming Meta Superintelligence as the newest Frontier Lab, >90% discount for traininglatent.space
“Google was ahead only a few hours”: Muse Spark 1.3 edges out Gemini as Meta claims its biggest coding leap yetthenewstack.io
Meta says Muse Spark 1.3 has frontier performance — but its best results come from a model developers can’t broadly use yetventurebeat.com
Meta AI Released Muse Spark 1.3: An Agentic Coding Model That Uses ~20% Fewer Tool Calls and ~25% Fewer Tokens Than Muse Spark 1.2marktechpost.com
- Sep 4
More stories today
Featured runs 100K monthly media pitches with 3 engineers on Vercel
Featured's agents deliver 100,000+ media pitches per month and have sent over 100 million HARO emails in the past year, connecting a user with a journalist every 6 seconds. The three-engineer team migrated 374 Sanity sites off AWS Elastic Beanstalk and runs its chatbot across 17 models via AI SDK and AI Gateway.
Vercel Blog·2 hours ago
WebBrain brings open-source AI agent to the browser sidebar
Robert Scoble·2 hours ago
OpenAI advises rewriting skills and AGENTS.md for GPT-6 Astra
OpenAI Developers·2 hours ago
ComicForge web app turns written stories into finished comics
TestingCatalog News·2 hours ago
AWS blog: choosing OpenAI models on Bedrock by outcome, not price per token
AWS argues dollars-per-million-tokens is the wrong comparison metric for production generative AI workloads, which buy outcomes like a resolved support ticket rather than tokens. The post covers picking among OpenAI models available on Amazon Bedrock.
AWS AI Blog·2 hours ago
Tailscale built customer-facing model router on Vercel AI Gateway
Aperture by Tailscale routes hundreds of AI models through Vercel AI Gateway and Sandbox, granting and revoking model access via tailnet network identity. Tailscale went from routing prototype to paying customers in months instead of building provider plumbing in-house.
Eclipse CEO Lior Susan urges AI industry collaboration over retreat
Eclipse CEO Lior Susan argues the AI industry should collaborate rather than pull back on advanced AI and data center expansion, warning a retreat risks ceding the US technological lead to China.
Bloomberg Technology·2 hours ago
Claude Code 2.1.269 release imminent
Claude Code Log·2 hours ago