SendTech Times
العربية
AINews|July 15, 2026, 05:50 AM
AI SHIFT:

Meta Opens Muse Spark 1.1 API With 1 Million-Token Agent Context

BySendTech Times AI & Enterprise DeskNewsroom-edited, source-reviewed coverage|Source: SiliconANGLE
Newsroom brief

SiliconANGLE reported that Meta launched Muse Spark 1.1 in Meta AI and a public-preview Meta Model API, with a 1 million-token context window and company-reported coding benchmark gains. The article did not report API pricing, named enterprise customers or independent performance validation.

Verified against source materialEdited by SendTech Times AI & Enterprise Desk
Meta Opens Muse Spark 1.1 API With 1 Million-Token Agent Context
Image source: SiliconANGLE

Muse Spark 1.1 is now available through the Meta AI chatbot and a public-preview Meta Model API, giving developers access to a new flagship Meta model for multi-agent automation workflows.

SiliconANGLE reported that the system is designed to keep more agent work inside context during long tasks.

The public-preview API lets developers embed the large language model in custom software.

Meta Model API Opens Muse Spark 1.1 To Developers

The new model is built for workflows where one main agent plans a task and subagents carry out parts of it.

According to SiliconANGLE, Meta said Muse Spark 1.1 can respond to mid-task developments that require the plan to change.

A context compaction mechanism handles data generated while agents work through multiple steps.

Meta said the mechanism preserves important details and lets the model retrieve information from earlier work when it needs to move data between sub-tasks, SiliconANGLE reported.

The context window for Muse Spark 1.1 is listed at 1 million tokens.

Context overflow was described as a quality risk when agents generate more data than the underlying model can retain.

Coding Benchmarks Remain Meta-Reported

Engineers at Meta tested Muse Spark 1.1 by asking it to generate a chat app from prompts, according to SiliconANGLE.

The model took screenshots of the interface, identified technical issues, found the code snippets behind them and fixed the problems.

According to SiliconANGLE, Meta said Muse Spark 1.1 scored 72.2 on Vibe Code Bench v1.1, an AI programming benchmark.

The same comparison put the model more than 50 points ahead of the company's previous flagship large language model, while a second test, SWE-Atlas Codebase QnA, showed a nearly 18% higher score.

Those benchmark claims remain company-provided.

SiliconANGLE did not report independent benchmark validation, customer deployment results or third-party testing for the new model.

Reuters Report Names Meta's 14-Gigawatt Capacity Plan

The report also cited a Reuters account that Meta plans to raise its data centre capacity to 14 gigawatts next year.

Reuters said the plan revolves around Iris, an internally developed AI chip set for mass production in September.

The same account said Iris is likely the MTIA400 chip that Meta previewed in March.

The chip includes 51% more high-bandwidth memory than Meta's previous-generation silicon and supports enhanced MX8 and MX4 data formats, according to SiliconANGLE.

Meta said MTIA400 is 400% faster than its predecessor, according to SiliconANGLE.

SiliconANGLE did not include pricing for the Meta Model API, named enterprise customers, independent performance validation or a production availability date beyond the public-preview API.

Share this article
inXf

Related articles

More
Anthropic Adds Nvidia BioNeMo Toolkit To Claude Science Beta
AI

Anthropic Adds Nvidia BioNeMo Toolkit To Claude Science Beta

Anthropic has launched a Claude Science public beta that integrates Nvidia BioNeMo Agent Toolkit. AI News cited 18 of the top 20 drugmakers using BioNeMo, but the companies did not disclose beta-user counts, named Claude Science customers or independent benchmark methodology.

Meta Pocket App Tests Prompt-Built Games In Social Feeds
AI

Meta Pocket App Tests Prompt-Built Games In Social Feeds

Meta has launched Pocket, an AI platform for making and sharing prompt-built mini games and interactive apps, according to AI Times Korea. The limited app-store rollout uses technology from the acquired Gizmo team, but Meta has not named a global launch timetable or creator monetisation terms.

Google Adds Gemini Agent To Search Ads In India Beta
AI

Google Adds Gemini Agent To Search Ads In India Beta

Google launched Business Agent for Leads in India as a Gemini-powered Search ad format that can chat with users on the search page. The company cited $77.25 billion in first-quarter ad revenue and several India ad metrics, but did not disclose pricing, wider rollout dates or independent lead-quality validation.

Bhavin Turakhia Backs Neo With $30 Mn For Enterprise AI Workflows
AI

Bhavin Turakhia Backs Neo With $30 Mn For Enterprise AI Workflows

Zeta founder Bhavin Turakhia has launched Neo with a $30 Mn personal investment, but the enterprise AI startup has not named paid customers, pricing or independent benchmarks.

Z.ai GLM-5.2 Pushes Open Coding Models Into Longer Workflows
AI

Z.ai GLM-5.2 Pushes Open Coding Models Into Longer Workflows

Z.ai released GLM-5.2 under an MIT license with a one million-token context window, coding-agent benchmarks and self-hosting options, putting long-context software engineering back into the open-model race.

MiniMax M3 turns long-context AI into an agent platform test
AI

MiniMax M3 turns long-context AI into an agent platform test

MiniMax launched M3 on June 1, 2026, combining long-context, agentic, coding and native multimodal capabilities in one model line. The API supports up to 1 million tokens of context, with a guaranteed minimum of 512K tokens, and includes M3 and M3-highspeed versions. MiniMax plans to open-source M3 on HuggingFace and GitHub, while early pricing offers a 50% discount for the first seven days.

Keep Reading

More Stories

Latest
Microsoft Revokes 11 Secure Boot Shims After ESET Finds Bypass RiskCybersecurityJul 15, 2026Microsoft Revokes 11 Secure Boot Shims After ESET Finds Bypass RiskArs Technica reported that ESET found 11 old UEFI shim images that Microsoft still trusted even after known defects. Microsoft revoked the shims in its June patch release, but the company has not explained how the lapse lasted for years.NVIDIA Lists Nemotron Enterprise AI Use Cases Without Contract DataAIJul 15, 2026NVIDIA Lists Nemotron Enterprise AI Use Cases Without Contract DataNVIDIA said its Nemotron open models are being customised by enterprise and national AI builders, with examples across clinical documentation, legal work, enterprise search and Malaysian-language AI. The company cited partner benchmark and cost claims, but did not disclose contract values, deployment volumes or independent benchmark audits.OpenAI Names GPT-5.6 For Microsoft 365 Copilot WorkflowsAIJul 15, 2026OpenAI Names GPT-5.6 For Microsoft 365 Copilot WorkflowsOpenAI said GPT-5.6 will power Microsoft 365 Copilot in Word, Excel, PowerPoint, Chat and Cowork. The official announcement names Microsoft API access and model-serving plans, but it does not give rollout dates, pricing changes, Copilot adoption metrics or customer benchmark results.OpenAI Researcher Disputes $2B AI Drug Startup ReportAIJul 15, 2026OpenAI Researcher Disputes $2B AI Drug Startup ReportTechCrunch reported that OpenAI researcher Miles Wang is leaving to form an AI drug-discovery startup and is in talks to raise about $200 million at a $2 billion valuation. Wang disputed the funding figures and company description, while the report did not include final deal terms, named co-founders or a launch date.Anthropic Commits $10 Million CAD To Canadian AI ResearchAIJul 15, 2026Anthropic Commits $10 Million CAD To Canadian AI ResearchAnthropic said it is committing $10 million CAD to eight Canadian research institutions and adding Canada’s three regional AI institutes to its startup programme. The company also published a country brief that ranked Canada eighth worldwide for Claude usage and second on a per-capita basis.New York Blocks New 50 MW Data Centre Permits For Up To A YearCloud & Data CentersJul 14, 2026New York Blocks New 50 MW Data Centre Permits For Up To A YearThe Verge reported that New York Governor Kathy Hochul signed an executive order blocking new environmental permits for hyperscale data centres above 50 megawatts for up to a year. The report said a stricter 20 megawatt legislative moratorium still awaits her decision.UMC Starts Singapore Silicon Photonics Production For AI NetworksChips & SemiconductorsJul 14, 2026UMC Starts Singapore Silicon Photonics Production For AI NetworksCNBC reported that United Microelectronics Corporation has started mass production of silicon photonics wafers in Singapore with SILITH Technology after an 18-month development push. The company plans a 12-inch platform for customer product development by 2027, while the report did not name first customers or supply commitments.Swiss Regulator Opens Google Android Choice-Screen ProbeCapital & PolicyJul 14, 2026Swiss Regulator Opens Google Android Choice-Screen ProbeTNW reported that Switzerland’s Competition Commission opened a preliminary investigation after Google removed the Android search-engine choice screen from Swiss phones while keeping it in the European Economic Area. COMCO said the change could affect rival search and digital-service providers, while Google has not published a rationale for the Swiss withdrawal.Anthropic Tokenizer Can Add Up To 73 Percent More Claude TokensAIJul 14, 2026Anthropic Tokenizer Can Add Up To 73 Percent More Claude TokensThe Register reported that Playcode found Anthropic's newer Claude tokenizer can emit up to 73 percent more tokens than OpenAI's GPT-5.x family on a TypeScript file. The same report said Anthropic's introductory Sonnet pricing runs through August 31, 2026, while independent customer-bill and task-completion evidence remains absent.NestAI Joins Finland-Estonia Defence AI Framework Without Budget CommitmentAIJul 14, 2026NestAI Joins Finland-Estonia Defence AI Framework Without Budget CommitmentTNW reported that Finland’s Ministry of Defence, the Estonian Defence Forces and Helsinki-based NestAI signed an artificial intelligence cooperation letter with no financial commitment. The framework covers joint development, training and technical cooperation, but it does not name procurement contracts, budgets or deployment dates.Meta Says Hyperion AI Campus Will Reach 5 GW In LouisianaCloud & Data CentersJul 14, 2026Meta Says Hyperion AI Campus Will Reach 5 GW In LouisianaData Center Knowledge reported that Meta's Hyperion AI campus in Louisiana is being expanded to 5 GW and more than $50 billion of investment, with Entergy filings listing 1,500 MW of new solar and hybrid resources and customer-funded transmission for the expansion.Cloudflare Precursor Scores Browser Sessions As Bot Traffic Hits 57 PercentCybersecurityJul 14, 2026Cloudflare Precursor Scores Browser Sessions As Bot Traffic Hits 57 PercentCloudflare made Precursor generally available to score visitor behaviour across full browser sessions rather than one arrival check. The public record still lacks pricing, customer adoption figures and customer false-positive rates for the session-scoring product.