Models & Labs · United States

Meta releases 816B-parameter Muse Spark Open for coding and agent work

Credited to SiliconANGLE · siliconangle.com · 20h ago

Photograph · SiliconANGLE

Meta Platforms Inc. today released open weights for its Muse Spark artificial intelligence model, which had been proprietary since its April debut.

The release, called Muse Spark Open, is built for coding and long-running agent work. Meta describes it as the most capable open-weight model to come out of a U.S. lab so far. Only 42 billion of the model’s 816 billion parameters switch on for any given token. Prompts can mix text with images, audio or video, and Meta has set the context window at up to 1 million tokens.

The starting point was Muse Spark 1.2, shipped in August, and Meta then added safety post-training so the model holds up without the guardrails of Meta’s own services. Meta said Muse Spark Open cleared the release thresholds in its Meta Superintelligence Scaling Framework, with a full evaluation report still to come. A technical writeup from Hugging Face Inc. says training leaned on what Meta calls “real developer workflows,” including multi-file refactors and long debugging sessions.

Meta also tested the model away from the data center, on a pair of Apple Inc. Mac Studio desktops with M5 Ultra chips that Apple’s MLX framework linked into one system. The two machines had about a terabyte of memory between them. Muse Code ran on that setup while the coding terminal worked through bug fixes and new features for the open-source Mastodon client Ice Cubes. Footage in DaVinci Resolve was the other test, and the model went through it frame by frame to pick out and crop the best shots for vertical video.

In Meta’s own benchmark table, Muse Spark Open scored 61.8 on the JobBench agent test, more than twice the 31.1 recorded by Inkling, the U.S. open model from Thinking Machines Lab Inc. Nvidia Corp.’s Nemotron 3 Ultra trailed further at 27.2.

Chinese open models are stronger at coding. Kimi K3 led the Terminal-Bench 2.1 coding test at 85, and GLM 5.3 and Qwen3.8 both finished above Muse Spark Open’s 76.4.

On tool use the order flips, and Muse Spark Open’s 89.2 on MCP-Atlas beat every rival Meta listed, the Chinese models included. The benchmark gives a model plain-language tasks without naming the tools needed to complete them. Working out which of 220 tools across 36 live Model Context Protocol servers to call, and in what order, is left to the model.

The weights come with a new Muse Model License carrying what Meta calls “limited commercial terms,” and no fee applies to developers who adapt the model or ship it inside their own software. Meta requires a separate agreement from any company that sells the model as a hosted service and has passed $20 million in gross revenue over a 12-month period. Products built on it have to show a “Built with Muse Spark Open” label. Military and critical infrastructure uses are off-limits.

Meta’s previous open-weight release was the much smaller Muse Glimmer, a 30 billion-parameter model that went out in August. The company’s most capable systems remain closed, and the Muse Wonder models introduced Monday are available only through the Meta Model API and Muse Code.

Seven inference providers offer hosted access at launch, Groq Inc. and Fireworks AI Inc. among them. The weights can also be downloaded from Hugging Face or from the model catalogs of Amazon Web Services Inc., Google LLC and Microsoft Corp.

Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities.

Are you an AWS customer? Support SiliconANGLE financially by buying your AWS services from our Marketplace portal page and links: https://siliconangle.com/aws-marketplace/

Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.

How agents are stoking new interest in a 40-year-old database

Vinci targets faster chip design and simulation with physics-driven AI

TSMC posts strong third-quarter results, but shares drop amid AI selloff

IBM ties intelligent operations to agent safety and business value

Nvidia expands Nemotron LLM family with hyper-efficient models for agentic systems

Startup nFox to open-source its 'sandbox' for open AI models

Join our community

Models & LabsUnited States

Tech, Media & Telecom Roundup: Market Talk

Credited to Barron's$ · wsj.com21h ago

Useful

Find insight on CoreWeave, Anthropic, the AI investment boom and more in the latest Market Talks covering Technology, Media and Telecom.

View the article · $Barron's

Front Method

Models & Labs · United States

Anthropic secured highest junk credit rating in June

Reading from TradingKey · Credited to Financial Times $ · ft.com · 14h ago

In June, KBRA assigned AI startup Anthropic a speculative-grade "BB+" credit rating with a stable outlook, just below investment grade. Despite incomplete financial data regarding free cash flow and EBITDA, the rating was supported by robust operational performance, including strong third-quarter revenue growth and consecutive adjusted profitability. However, slowing growth rates and massive computing power investments highlight long-term financing and cash flow risks. Achieving a future investment-grade rating would lower capital costs and broaden access to long-term institutional funds like pension funds, aligning with similar pre-IPO financial strategies pursued by rival OpenAI.

TradingKey - Anthropic's Credit Rating Surfaces for the First Time.

According to documents obtained by the Financial Times, credit rating agency Kroll Bond Rating Agency (KBRA) assigned Anthropic a "BB+" rating with a stable outlook in June this year. This is the highest tier among speculative-grade ratings, just one notch below the investment-grade threshold.

For an AI startup founded just five years ago with relatively limited financial disclosures, this outcome is not bad. If Anthropic achieves an investment-grade rating in the future, the company will gain access to larger pools of long-term capital, such as life insurers and pension funds, while reducing financing costs for debt issuance and data center construction.

The rating was established prior to the closing of a $35 billion chip financing deal. The financing package was led by Apollo Global Management ( APO ), with participation from BlackRock ( BLK ) and several banks, primarily to support Broadcom's ( AVGO ) AI chip platform. Of this, approximately $4.4 billion in subordinated debt directly exposes investors to Anthropic's credit risk, while the remainder is supported by Broadcom. The issuance yield on the subordinated debt was around 8.5%, while the interest rate on the senior debt was 5.75%.

Notably, KBRA assigned the rating despite having incomplete financial data.

Anthropic provided the rating agency with revenue, adjusted operating profit, annualized first-quarter revenue, and second-quarter internal forecasts, but did not disclose key metrics such as free cash flow and EBITDA. Analysts could only estimate the company's debt-service capability based on limited data and their own assumptions, employing a risk-scoring approach instead of a traditional comprehensive financial model.

Despite disclosure gaps, Anthropic's recent operating performance provided support for its credit rating. The company stated at a pre-IPO investor presentation that third-quarter revenue grew approximately 46% quarter-over-quarter to $16.8 billion, marking its second consecutive quarter of adjusted profitability. Actual second-quarter revenue was approximately $11.5 billion, higher than the internal forecast of $10.9 billion it previously provided to KBRA.

However, Anthropic's growth rate has slowed noticeably compared to the beginning of the year, and its third-quarter results fell short of targets previously set by some optimistic investors. This implies that while the company is improving profitability, it has yet to fully demonstrate that its revenue growth can cover its massive computing power investments over the long term.

As of the end of June, Anthropic carried no debt on its own balance sheet. In the past, the company primarily relied on commercial partners with higher credit quality and special purpose vehicles to raise funds for chip procurement and compute leasing. While this model temporarily alleviates on-balance-sheet debt pressure, it cannot eliminate the financing risks ultimately borne by Anthropic's lease payments and operational cash flow.

Anthropic is not the only AI company seeking higher credit ratings; rival OpenAI is also preparing for a potential IPO, with financial advisors for both firms lobbying major rating agencies for investment-grade ratings post-listing.

This content was translated using AI and reviewed for clarity. It is for informational purposes only.

Comments ( 0 )

Click the $ button, enter the symbol, and select to link a stock, ETF, or other ticker.