A Product Hunt listing blowing up right now is all about Meta's latest bet: Muse Spark 1.1, released by Meta Superintelligence Labs on July 9. This is the company's first serious attempt to enter the paid API market for advanced models — and they're not coming in with their hat in hand.

What's getting people excited isn't necessarily the performance in isolation. It's the price combined with what the model is actually built for. Muse Spark 1.1 costs $1.25 per million input tokens and $4.25 per million output tokens. By comparison, Claude Opus 4.8 sits at $5/$25 and GPT-5.5 at roughly $5/$30. That's a meaningful price difference when you're running real agentic workflows in production.

Saoud Rizwan from Cline says Meta is "clearly building for serious agentic coding at a price point that makes it viable to run real coding jobs at scale".

The model itself is designed to orchestrate multi-agent systems — meaning not just answering questions, but planning, delegating to subagents, and completing long-horizon tasks across applications. It handles text, image, and audio as input, and has context compression built in to keep track of critical steps during long sessions.

But — and this is important to note — Muse Spark 1.1 is not the sharpest tool in the shed on pure coding benchmarks. On Terminal-Bench 2.1 it scores 80.0, while Claude Opus 4.8 sits at 82.7 and GPT-5.5 at 83.4. Meta itself acknowledges that long-horizon agentic work still lags slightly behind the two top competitors. Amjad Masad from Replit highlights frontend and design as a strength, but not everything.

Meta crashes the party: Muse Spark 1.1 challenges GPT-5.5 at half the price - Bilde 1

What's interesting here is the strategy: Meta isn't trying to win every benchmark. They're going after being good enough on agentic tasks, and cheap enough that it actually makes sense to build on them in production. It's a clear signal that Meta wants a foothold in the enterprise and developer market — not just in consumer AI.

The source is community-based and early, so take the numbers with a grain of salt until third-party tests start emerging. But the signal is clear enough: Meta is no longer just a spectator in the frontier model race.

Free access is available in the Meta AI app ("Thinking" mode), and the API is in public preview in the US with $20 in free credits for new accounts. Worth testing if you're working with agentic pipelines.