← All benchmarks

B-Tier

Muse Spark 1.2

Meta · Open-weight generalist, early evidence

Meta's current open-weight release — promising, but still building its track record.

Benchmark runs not yet completed

Why this tier

Muse Spark 1.2 is marked 'Full info' rather than a run count, meaning we have a model profile but haven't yet put it through our full repeated-trial suite. Early signal places it in B-tier — a capable open-weight generalist — but we're holding off on a firmer verdict until it's been through the same ten-run process as its tier-mates.

Strengths

  • Open licensing, straightforward to self-host and fine-tune
  • Backed by Meta's long open-weight research track record
  • Reasonable general-purpose capability from early testing

Worth knowing

  • Not yet through our full ten-run benchmark suite — treat current tier as provisional
  • Open-weight deployment still requires real infrastructure investment

Best for

  • Teams already building on the Llama/Meta open-weight ecosystem
  • Fine-tuning and customisation projects
  • Early evaluation before a larger commitment

Not sure which model fits?

We'll help you pick the right one for the job.

Model choice is rarely one-size-fits-all. Tell us what you're building and we'll tell you plainly what fits.