Meta has released pricing details for its Muse Wonder model, with input tokens costing $4 per million and output tokens at $20 per million. Caching can reduce input costs to $0.20 per million tokens, and reasoning tokens are billed as output. While the per-token cost is higher than its predecessor, Muse Spark 1.3, the model offers improved performance on benchmarks. However, the cost per completed task can be significantly higher due to token volume, especially in agentic loops where context is repeatedly sent. Muse Wonder Full is available through Muse Code subscriptions, not via a public API. AI
IMPACT Sets a new pricing tier for frontier models, potentially influencing competitor pricing and enterprise adoption strategies.
RANK_REASON Pricing details for a new model from a major AI developer.
- Anthropic
- Meta
- Muse Wonder
- Opus 5.5
- Artificial Analysis
- .claude
- GPT-6 Astra
- Meta Model API
- Muse Code
- Muse Spark 1.3
- Muse Wonder (Full)
- .openai
AI-generated summary · Google Gemini · from 6 sources. How we write summaries →