"Muse Spark Open nuked a city for 10x less than Muse Wonder" (a voxel city with TNT)
DesignRead the post
x.com · link re-checked 2026-10-16
Muse Spark
By Meta. 3 results on 3 boards, copied as published. Different boards, different metrics: don't add them up.
License (as listed)Not listed
Hosting regionNot confirmed
Boards3
Code
Agentic / terminal
No result on the boards we captured
Websites
- 19.67%Vals AI Vibe Code Bench v1.1
Muse Spark · Score % · board #87
OpenHands · as of 2026-10-20 ↗
Design
No result on the boards we captured
Android
No result on the boards we captured
iOS
No mature public leaderboard
Available in
Muse Code (Meta) Everyday Usage→ Muse Code (Meta) High Usage→ Muse Code (Meta) Power Usage→ Muse Code (Meta) Maximum Usage→
Use cases with this model
Launch: Muse Spark 1.3 "frontier performance almost too cheap to meter … Try it in Muse Code"
CodeRead the post
x.com · link re-checked 2026-10-16
Same family: Muse 15
Llama: only local infra (e.g., Llama 3.2 3B at 124 tok/s on a jailbroken PS5)
CodeRead the post
x.com · link re-checked 2026-10-16
Muse CodeMuse family
Unofficial Muse Code desktop app built on top of the Muse Code CLI
CodeRead the post
x.com · link re-checked 2026-10-16
Muse CodeMuse family
Complaint: "Meta muse code is only available on US or canada"
CodeRead the post
x.com · link re-checked 2026-10-16
Test: "Muse Wonder Full 1.0 on real work. 2 repos. 105 bugs … GPT-6 Astra (max): 45 Muse Wonder Full 1.0 (max): 44 Fable 5.1 (max): 43 Opus 5.5…
CodeRead the post
x.com · link re-checked 2026-10-16
Test: the same harness on plain Muse Wonder 1.0 scored "31.7", vs "Muse Spark 1.3 (max): 32.2"
CodeRead the post
x.com · link re-checked 2026-10-16
Test: an animated Popeye website from scratch, Muse Wonder vs Opus 5.5
WebsitesDesignRead the post
x.com · link re-checked 2026-10-16
Sources
vals.ai/benchmarks/vibe-code vals.ai/benchmarks/swebench scale.com/leaderboard/swe_bench_pro_public