What happened
Meta opened Muse Spark 1.1 to developers through a new Model API. The company lists the model at $4.25 per million output tokens and is pitching it for agentic workloads, bug fixing, and code migrations.
The move places Meta in a coding-model market where pricing and performance-per-dollar claims are becoming a central competitive point. The upstream record describes the listed price as a fraction of OpenAI or Anthropic list prices, but does not provide those comparison prices.
Why it matters
A lower-priced coding option could add pressure to the economics of frontier AI labs and give buyers more leverage. The related record also notes that Databricks switched to open-source GLM 5.2 for cheaper coding, pointing to broader demand for lower-cost alternatives.
Price alone does not establish which model will perform best for a particular team or task. The record says public coding benchmarks used in these comparisons are contested.
What to watch next
Watch for independent, workload-specific testing of Muse Spark 1.1, especially on the bug-fixing, code-migration, and agentic tasks Meta highlights. That is the more useful receipt than a broad benchmark comparison.
Simba Pool publishes this quick AI brief from supplied upstream research and does not provide investment advice.
Watch for independent tests of Muse Spark 1.1 on real coding workloads and for changes in competing model prices.
Upstream references
Digest dated 2026-07-10 · upstream model claude-sonnet-4-6. Source IDs are preserved for audit; the publishing host does not receive the upstream URL map.
- 1
62f0f89ded5a3e40cb6f217de19f6e2a0ffc1dd0Reference from the upstream research server - 2
94cae6cec825c1fb519b39460c3a7028920bfeadReference from the upstream research server - 3
57b32d30c2acccf9caab2264cca84dffe28f8191Reference from the upstream research server
This quick brief was generated by Terra from a dated upstream research digest. It has not received the source-by-source human review required for a Reviewed analysis. Material limit: The record confirms Meta's launch and listed price, but it does not provide independent performance results; public coding benchmarks are contested.