- Bay Area Times
- Posts
- Meta debuts closed Muse Spark, breaks from open Llama as first Wang-led MSL model
Meta debuts closed Muse Spark, breaks from open Llama as first Wang-led MSL model
Bay Area Times
April 09, 2026
Brought to you by:
Top stories today:
- Meta debuts closed Muse Spark, breaks from open Llama
- Claude Managed Agents launched to build, deploy AI agents at scale
- xAI reorganizes engineering as SpaceX SVP Nicolls becomes president
- Google’s Gemini expands NotebookLM integration with “Notebooks”
- OpenAI Foundation to finalize $100M+ Alzheimer’s grants
- Nebius explores AI21 Labs deal after NVIDIA’s $3B talks failed
0. Data and calendar
All values as of 6 AM ET / 3 AM PT, other than S&P500 and NASDAQ close (4 PM ET / 1 PM PT).
All times are ET.
Listen to our AI-generated podcast summarizing today’s newsletter (beware of hallucinations):
1. Meta debuts closed Muse Spark, breaks from open Llama as first Wang-led MSL model
Muse Spark is just behind Gemini 3.1 Pro, GPT-5.4, and Claude Opus 4.6 on the Artificial Analysis Intelligence Index:
Meta’s first release since Llama 4 in Apr. 2025.
No public API is available yet.
Meta is opening a private API preview for select partners and plans to offer paid API access to a wider audience later.
Powering Meta AI across Facebook, Instagram, and Threads.
An open-source Muse Spark version is in the plans.
Meta shares closed +7% following the announcement.
Muse Spark used 58M output tokens on the Intelligence Index, vs. Gemini 3.1 Pro (57M), Claude Opus 4.6 (157M), and GPT-5.4 (120M)
Stop swapping between tools. 150,000+ Microsoft customers use Slack to connect every tool and automate workflows.
Watch the demo to see integrations with Office and SharePoint, AI-powered automation, and no-code search.
Benefit from an open ecosystem with 2,600+ apps that puts AI at the center of collaboration.
*Sponsored.
3. Claude Managed Agents launched by Anthropic to build, deploy AI agents at scale
Managed Agents lets businesses leverage Anthropic’s infrastructure to deploy agents at scale.
Developers get an agent harness, the infrastructure layer that enables models to operate agentically.
The tool includes a sandboxed environment for agents to spin up software projects securely.
Developers can run agents autonomously for hours in the cloud and monitor other agents.
Managed Agents is available in public beta on the Claude platform.
Separately, an appeals court denied Anthropic’s bid to pause the DoD risk label
Acting Attorney General Todd Blanche hailed the ruling as a victory for military readiness:
A panel of the U.S. Court of Appeals denied the bid to pause the designation during the case.
The decision is not final.
Anthropic said it is confident the court will ultimately rule the designation unlawful.
Mar. 26: a California federal judge blocked one of the orders designating Anthropic under two laws.
Additionally, Anthropic completed an employee tender at a $350B valuation, below the $6B sought as staff held shares.
4. xAI reorganizes engineering as SpaceX SVP Nicolls becomes president, says it is “clearly behind”
