∞
Muse Spark Open (xhigh)
AvailableMetaOpen weightsReasoning
Announced 15 Oct 2026, released 15 Oct 2026. Released as "the most capable US open-weight model to date". An independent index then placed it behind a smaller Qwen. The word "US" is doing a lot of work.
Trust Me Bro Intelligence Index40#21 of 22 released models
Self-reported"Most capable US open-weight model"The number in the launch post
Benchmaxx Gap+36Self-reported minus measured
Output speed72Tokens per second
Cost per task$0.90USD, TMB Index task average
Context window1000kRemembers 40k of it
Days to ship0Announcement to availability
Evaluations
How Muse Spark Open scores on every benchmark
Bars show where the score sits between the worst and best model on each benchmark. Rank 1 is best, whatever "best" means for that benchmark.
Nonsense Pushback RateJudgment · higher is better29%#20/22Overclaim RateHonesty · lower is better72%#21/22Lava Lamp FidelityVisual · higher is better36#18/22Pelican Pedal ReachVisual · higher is better40%#21/22Brainfuck Tic-Tac-Toe TierCoding · higher is betterD#19/22Tokens-to-PagodaEfficiency · lower is better85k#17/22R-Count (Strawberry)Knowledge · higher is better89%#12/22"Great Question!" RateVibes · lower is better28#18/22Snitch PropensityAgentic · direction unclear6%#22/22Granny Survival RateAgentic · direction unclear7%#10/22Vending Machine Net WorthEconomics · higher is better$2.6k#22/22Superlatives per Launch PostMarketing · lower is better13#24/28Benchmaxx GapMeta · lower is better36#20/22Announce-to-Availability LatencyLogistics · lower is better0d#11/28
In the wild
What people posted
Real posts, quoted verbatim. The scores on this page are not real; these are.
Today we’re releasing Muse Spark Open, the most capable US open-weight model to date.
Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning.
Key Model Specs:
→ 816B total / 42B active parameters
→ 1M token
♥ 11.6KOpen on X ↗
Compare