Meta has officially unveiled a major update to its neural network Muse Spark 1.1, and this event is radically reshaping the landscape of language models. For the first time, the company is bringing a commercial AI model to market, with an aggressive pricing strategy that threatens the positions of Anthropic and OpenAI.
Aggressive Pricing as the Main Weapon
The cost of using the Muse Spark 1.1 API is set at a record low. Processing 1 million input tokens will cost just $1.25, while generating 1 million output tokens will cost $4.25. For comparison: Anthropic's base model Sonnet 5 charges $2 for input and $10 for output, while the flagship Opus 4.8 and GPT-5.5 cost $5 and $5/$30 respectively.
Thus, Muse Spark 1.1 is 37% cheaper than Sonnet 5 for input tokens and 75% more affordable than Opus 4.8 and GPT-5.5. The gap for output tokens is even more impressive: $4.25 versus $10 for Sonnet 5, $25 for Opus, and $30 for GPT-5.5. This is a direct attack on developers' wallets, especially in the code writing and autonomous AI agent creation segments, where Anthropic and OpenAI products dominate.
Does Muse Spark Have Real Chances?
According to early testers, Muse Spark 1.1 demonstrates performance comparable to GPT-5.5 and Claude Opus 4.8 in complex autonomous execution tasks. However, Meta has not yet provided independent benchmarks, and access to the platform is strictly limited: only developers from the US, after application approval, can evaluate the functionality in preview mode.
The low entry threshold and a generous $20 testing bonus for newcomers are powerful incentives for the IT community to migrate. If the model's quality truly matches the claims, we will witness a serious reshuffling of the commercial AI market.
My opinion: Meta is deliberately sacrificing margins to capture market share, leveraging its infrastructure power. This is a classic "strangle" tactic of competing on price. If Muse Spark proves its effectiveness, OpenAI and Anthropic will either have to sharply reduce their rates or prove that their premium quality is worth a 3-5x premium. For developers, this is unequivocally a positive signal — the AI market is becoming more accessible.