Meta's unannounced Muse-Spark-1.4 model spotted in GitHub PR with real API usage logs
testingcatalog · x · 2026-11-14
A model ID muse-spark-1.4-internal was discovered in a Meta PR on GitHub — a small logging change to RocksDB's experimental live tuning feature, with the mention appearing in auto-generated PR notes.
The screenshot shows an actual model call with detailed usage: 33,535 input tokens, 17,780 cached tokens, 8,169 output tokens, 7,867 reasoning tokens, in 64.96 seconds — indicating the model is already in internal use at Meta. Muse Spark 1.4 appears to be in development and deployed internally.
More from Models
- The Newest LLM Isn't Automatically the Best Business Choice — DavidLinthicum · 2026-11-14
- XERJ Reader-1 launches in preview: verifiable document Q&A at $0.01 per 1M input tokens — omarsar0 · 2026-11-14
- ChatGPT now shows disclaimers in replies and cites sources by name — lilyraynyc · 2026-11-14
- Pawel Huryn charts cost-per-task across models and effort levels — PawelHuryn · 2026-11-14
- Static model routers fail on long tasks — routing must be a dynamic handoff policy, dev argues — petrusenko_max · 2026-11-14
- Cloudflare hosts Thinking Machines' inkling 2 small on Workers AI at day-0 speed — michellechen · 2026-11-14