Meta ships Muse Spark 1.1 with a cheap Model API - the third-runner squeeze play

Meta released Muse Spark 1.1 (multimodal reasoning) and simultaneously priced its Meta Model API aggressively low. Neither will beat GPT-5.6 or Claude on benchmarks. That's not the point. The point is undercutting the meter.

The fact

Meta released Muse Spark 1.1, a multimodal reasoning model, alongside a cut-price Meta Model API (ITmedia). Pricing details put it well below OpenAI and Anthropic's equivalent tiers.

Our reading

Meta is playing the third-runner strategy: not lead the frontier, but drag the price floor down and force the frontier labs to defend their per-token economics. It worked for Llama; it's the same play with Muse Spark's paid tier.

The moment matters because it lands the same week OpenAI cut GPT-5.6's price and reset ChatGPT usage limits. The market message from three vendors in one week: AI inference is racing to commodity pricing - and the only sustainable moats will be data, distribution, and vertical integration into the enterprise workflow.

What to watch

Whether Meta open-weights Muse Spark's smaller variants (base rate for a Meta model). If yes, downstream cost pressure on OpenAI and Anthropic gets worse. If no, the "third runner" thesis stalls and Meta becomes another walled-garden model vendor.