Meta Launches Muse Spark 1.3 Max with Major Coding and Agent Gains
On September 5, Meta's Chief AI Officer and Scale AI founder Alexandr Wang publicly released Muse Spark 1.3 Max, with reasoning level set to the maximum tier and safety testing completed before launch. He said the new version is markedly stronger at coding and agentic tasks, and strongly recommended that users who had tried 1.3 High or XHigh try Max again. The release came only about two days after version 1.3.
Confirmed
- The model is available via the Muse Code CLI tool and the Meta Model API, with a curl install command provided by Wang for direct terminal installation.
- Meta's official account announced the max reasoning mode, pitched as "frontier performance, non-frontier pricing."
- Meta's Superintelligence Labs showcased demos including 3D content generation with the model; Wang reshared it saying "We're so back."
- Resharer @baaadas highlighted progress in visual coding.
- Per a third-party comparison relayed by @Lianhuiq, rebuilding a 3D model of Helsinki Cathedral across 1.1→1.2→1.3→1.3 Max showed a clear leap in detail fidelity for 1.3 Max, with lower total generation cost.
- On September 6, user @tinkerersanky praised the model as Opus-tier; Wang reshared and recommended trying it, an official-side amplification of positive word of mouth.
Why it matters
- This is Meta's latest move in the coding and agent race, directly targeting competitors known for strong reasoning and coding, and already benchmarked by users against Claude Opus.
- The pricing emphasizes cost-effectiveness, with third-party tests showing better detail at lower total cost, potentially intensifying price competition among frontier models.
- Shipping a Max tier only two days after 1.3, plus repeated official recommendations to High/XHigh users, suggests the reasoning gains are seen as a generational improvement and reflects Meta's rapid iteration pace.
2026-09-05 ~ 2026-09-06 · 12 related posts
- Episode 1: Meta Launches Muse Spark 1.3: Near-Frontier Performance at ~1/8 the Price(2026-09-03, 53 posts)
- Episode 2: Meta Releases Muse Spark 1.3 as Its Strongest Model(2026-09-03, 2 posts)
- Episode 3: Meta's Muse Spark 1.3 Nears Fable 5.1 in Community Benchmarks at Fraction of Cost(2026-09-03, 5 posts)
- Episode 4: Meta Chief AI Officer Praises muse spark 1.3 Sub-Agent Demo(2026-09-03, 2 posts)
- Episode 5: Meta Muse Spark 1.3 Ranks Second on Agentic Benchmark at a Third of the Cost(2026-09-04, 2 posts)
- Episode 6: Meta Launches Muse Spark 1.3 Max with Major Coding and Agent Gains(2026-09-05, 12 posts)
Primary sources
- Alexandr Wang Announces Muse Spark 1.3 Max With Stronger Coding and Agentic Performance — alexandr_wang ·
- Alexandr Wang endorses Muse Spark 1.3 Max for its 3D generation skills — alexandr_wang ·
- Meta ships Muse Spark 1.3 with max reasoning, pitching frontier performance at non-frontier prices — AIatMeta ·
- Muse Spark 1.3 max released with significantly stronger coding and agentic performance — alexandr_wang · 2026-09-05
- Alexandr Wang teases Muse Spark 1.3 with a max reasoning level after safety testing — alexandr_wang · 2026-09-05
- [source] Meta ships Muse Spark 1.3 with max reasoning, pitching frontier performance at non-frontier prices — AIatMeta · 2026-09-05
- Alexandr Wang Announces Muse Spark 1.3 with Max Reasoning and One-Line CLI Install — alexandr_wang · 2026-09-05
- [source] Alexandr Wang endorses Muse Spark 1.3 Max for its 3D generation skills — alexandr_wang · 2026-09-05
- Meta publicly releases Muse Spark 1.3 max with stronger coding and agentic performance — EdwardSun0909 · 2026-09-05
- Meta ships Muse Spark 1.3 Max in two days: sharper 3D detail at lower cost — Lianhuiq · 2026-09-05
- Alexandr Wang urges users to try Muse Spark 1.3 max, called 'Opus-level' by users — alexandr_wang · 2026-09-06
4 near-duplicate retellings: alexandr_wang · jordihays · alexandr_wang · baaadas