Meta Opens Its Frontier Model to Developers for the First Time
Muse Spark 1.1 is Meta's second model from Meta Superintelligence Labs — and the first Meta frontier model developers can build on, via a new paid, OpenAI-compatible API.
Meta has released Muse Spark 1.1, the second model out of Meta Superintelligence Labs, and — for the first time — opened it to outside developers through a new Meta Model API, now in public preview for U.S.-based developers.
What's new
Muse Spark 1.1 upgrades the original Muse Spark model that debuted in April. It's a multimodal reasoning model built for agentic tasks, with:
- A 1-million-token context window
- Improved tool use, computer use, and coding performance
- Stronger multimodal understanding
It's also available directly to consumers in "Thinking" mode inside the Meta AI app and on meta.ai.
The API is the real news
Until now, Meta's frontier models were closed to outside developers. The new Meta Model API changes that — and it's Meta's first-ever paid developer API:
- Pricing: $1.25 per million input tokens, $4.25 per million output tokens
- Format: OpenAI-compatible, so existing code built against the OpenAI SDK can point at Meta's endpoint with minimal changes
- Onboarding: new developers get $20 in free credits before moving to pay-as-you-go billing
- Access: public preview, U.S.-based developers only, as of launch
An "OpenAI-compatible" API means it accepts the same request format OpenAI's API uses — so developers don't have to rewrite their integration code to try a different model provider.
Why it matters
For developers: a third major U.S. lab (alongside OpenAI and Anthropic) now offers a commercial, general-purpose frontier API — more competition on agentic and coding tasks specifically, an area every lab is racing to own.
For enterprises: Meta entering the paid API market signals it intends to compete for the same agentic-workflow and coding-assistant budgets that OpenAI's Codex tooling and Anthropic's Claude Code already target.
For the broader market: OpenAI-compatible pricing at $1.25/$4.25 per million tokens puts Meta in direct competitive range with mid-tier offerings from its rivals, rather than trying to win purely on price.
What to watch
- Whether Meta expands the API beyond U.S. developers and public preview
- How Muse Spark 1.1 benchmarks against Claude and GPT-5.6 on independent agentic and coding evaluations
- Whether Meta's paid-API move changes its historically open-weight-first model strategy going forward
Source: Meta AI
NextGen AI Digest Editorial
Editorial Team
Reporting and analysis from the NextGen AI Digest newsroom — covering AI, agentic systems, SaaS, and the future of technology. Every piece is factual, sourced, and cited. Built and published by the team at Peaders.
Keep reading
Meta Enters the AI Coding Agent Race With Muse Code
Meta launched its first AI coding agent, Muse Code, alongside an upgraded Muse Spark 1.2 model — a direct challenge to OpenAI's Codex and Anthropic's Claude Code.
MCP Startup Runlayer Sues Rippling Over Gateway Product
Runlayer alleges Rippling evaluated its MCP gateway product, then built a competing version in-house — a dispute that lands squarely in the fast-moving agentic AI infrastructure space.
Microsoft Launches Its First Cybersecurity AI Model
Microsoft has launched its first dedicated AI security model alongside a new agentic cybersecurity system, bolstering its AI security offerings.