Meta Unveils Muse Spark 1.1, Opens Developer Preview

Meta released Muse Spark 1.1 in a U.S. public preview on the Meta Model API, offering new developers $20 in credits and pricing at $1.25 per million input tokens and $4.25 per million output tokens. The model targets agentic workflows, repository-level coding, and multimodal inputs, positioning it as a lower-cost competitor for production AI assistants.

Meta Unveils Muse Spark 1.1, Opens Developer Preview Meta released Muse Spark 1.1 in a U.S. public preview on the Meta Model API, giving new developers $20 in credits and listing $1.25 per million input tokens and $4.25 per million output tokens. Meta's developer materials frame the model around agentic workflows, repository-level coding, and multimodal inputs, while Reuters and Business Insider report Meta is positioning it as a lower-cost competitor for production assistants. The practitioner implication is cost-aware agent design: teams can test broader tool-calling and long-context workflows, but they still need independent benchmarks for code-edit reliability, latency, and safety before shifting high-volume workloads. Lower model pricing is most useful when the model can reliably hold state, call tools, and complete repository-level work. The LDS takeaway is that Muse Spark 1.1 should be evaluated as an agent platform input, not just as another chat model, because its economics only matter if validation and orchestration costs stay controlled. What happened Meta made Muse Spark 1.1 available to U.S. developers through the Meta Model API, according to Meta developer materials and Reuters. The developer materials list $20 in starter credits for new accounts and pay-as-you-go pricing of $1.25 per million input tokens and $4.25 per million output tokens. Business Insider also reported Meta's cost-positioning comments around the launch. Technical context Meta's materials describe the model around agentic workflows, advanced coding, and multimodal perception. For engineering teams, those claims map to three measurable questions: whether the model can plan and recover across tool calls, whether it can make repository-level edits without brittle context handling, and whether multimodal inputs improve workflows rather than adding new validation cost. For practitioners Teams evaluating the preview should run side-by-side tests on real tickets, code-search tasks, document-heavy triage, and tool-use workflows. The published token prices make experimentation easier, but total cost of ownership will depend on retry rates, latency, guardrail overhead, and how much human review is still needed. What to watch Independent agent and coding benchmarks will matter more than launch claims. Also watch billing behavior, rate limits, and SDK compatibility as teams move from experiments to long-running agents. Key Points - 1Meta priced Muse Spark 1.1 around mid-tier production agents, with starter credits and published per-token rates for developers. - 2The model targets agentic workflows, repository-scale coding, and multimodal inputs, making evaluation quality as important as list price. - 3Practitioners should benchmark tool-calling reliability, latency, and safety controls before moving high-volume automation to the new API. Scoring Rationale Meta opening a priced developer preview for an agent-oriented, multimodal model is notable because it changes cost and platform choices for production AI assistants. The impact is major enough for practitioners to track, but still depends on independent benchmarks and real-world reliability rather than launch claims alone. Sources Public references used for this report. 01ai.meta.comIntroducing Muse Spark 1.1 https://ai.meta.com/blog/introducing-muse-spark-meta-model-api/ 02developer.meta.comBuild with Muse Spark, now available on Meta Model API https://developer.meta.com/ai/resources/blog/build-with-muse-spark/ 03dev.meta.aiGet started with Meta Model API https://dev.meta.ai/docs/getting-started/overview/ View 8 more sources 04Meta debuts Muse Spark 1.1 model with preview open to developersreuters.com https://www.reuters.com/business/meta-debuts-muse-spark-11-with-preview-open-developers-2026-07-09/ 05Meta launches Muse Spark 1.1, promises cost-effective AIbusinessinsider.com https://www.businessinsider.com/meta-launches-muse-spark-1-1-cost-effective-ai-2026-7 06Meta jumps into AI coding market in effort to chase Anthropic and OpenAIcnbc.com https://www.cnbc.com/2026/07/09/meta-jumps-into-ai-coding-market-to-chase-anthropic-and-openai.html 07Meta says its new AI model is ready to compete on codingtheverge.com https://www.theverge.com/ai-artificial-intelligence/963193/meta-muse-spark-model-api 08Meta updates its Spark model, releases developer versionaxios.com https://www.axios.com/2026/07/09/meta-ai-spark-model-update-developer 09Muse Spark 1.1: Meta's Agentic Model and API - DataCampdatacamp.com https://www.datacamp.com/blog/muse-spark-1-1 10Muse Spark 1.1 vs Grok 4.5: Which Agent Model Wins?digitalapplied.com https://www.digitalapplied.com/blog/muse-spark-1-1-vs-grok-4-5-agentic-model-comparison-2026 11Meta prices Muse Spark 1.1 API at $1.25/$4.25 per M tokens | AI ...aiweekly.co https://aiweekly.co/alerts/meta-prices-muse-spark-11-api-at-125425-per-m-tokens Practice with real Ad Tech data 90 SQL & Python problems · 15 industry datasets Active Search Campaigns by BudgetEasy /problems/sql/active-search-campaigns-by-budget High CPC Clicks & Poor Landing PagesMedium /problems/sql/high-cpc-clicks-poor-landing-page Campaign ROAS by Attribution ModelHard /problems/sql/campaign-roas-by-attribution-model 250 free problems · No credit card See all Ad Tech problems /problems/datasets/adtech