← The Polylog AI Intelligence Brief
Morning Edition · Tuesday, July 14, 2026Published at 1:33 AM EDT · New York
Meta Opens a Paid Model API With Muse Spark 1.1, Its Turn Toward Metered Access
The Superintelligence Labs model ships a one-million-token context window and agentic tool use, priced at $1.25 per million input tokens and $4.25 per million output tokens.
Meta released Muse Spark 1.1 on July 9. For the first time, it made a first-party model available only through a paid, metered endpoint called the Meta Model application programming interface (API), now in public preview for developers in the United States. The company describes Muse Spark 1.1 as a multimodal reasoning model built for agentic tasks, with a one-million-token context window and improvements in tool use, computer use, coding, and multimodal understanding. It is the second model from Meta Superintelligence Labs and an upgrade of the original Muse Spark that debuted in April.
Pricing is what makes this release significant. Meta lists $1.25 per million input tokens and $4.25 per million output tokens, with $20 in free credits for new sign-ups before pay-as-you-go rates apply. That places the model in direct price competition with the metered APIs from OpenAI, Anthropic, and Google, rather than in the download-and-self-host approach Meta established with Llama.
The capability claims come from a vendor launch, and independent reproductions of the tool-use, coding, and long-context benchmarks are not yet available. What is verified is the commercial structure. There are no downloadable weights, there is a published rate card, and the preview is limited to one region.
Veracity: Corroborated
88/100- If true, who benefits
Closed-API incumbents gain a well-capitalized price rival, and Meta shifts revenue toward inference billing, while the framing pressures anyone standardized on downloadable Llama weights.
- The nuance
The commercial structure (paid US-only endpoint, $1.25 and $4.25 per million tokens) is independently confirmed, but the "retreat from open weights" is a narrative Meta has not stated, and every capability benchmark remains vendor-supplied and unreproduced.
Townsquare
Start a discussion in Townsquare.
More from this edition
- Anthropic Restores Fable 5 After a Cyber Jailbreak Forced It Offline
- Meta's Non-Invasive Brain-to-Text Decoder Reaches 78% Word Accuracy for Its Best Subject
- Bilibili Releases Index-1.9B, a Small Open Model Trained on 2.8 Trillion Tokens
- Researchers Bypass MedGemma-4B's Safety Rules With Trivial Prompt Reframing
- Meta Ships Muse Image and Muse Video With Editable, Agentic Generation
- Paper Shows Prompt Formatting Alone Can Flip Language-Model Leaderboards
- Japan's JX Metals Reports 90% Lithium Recovery From Spent EV Batteries
- Study Finds Message Format Effects in Multi-Agent Handoffs Depend on Model Tier
Comments
0No comments yet.