What Changed This Snapshot?

Note, added November 2026: This post was written before we fixed how benchmark statements are grouped. Until 21 October 2026, statements that read alike were merged even when they were about different models, tasks or snapshots, so a figure here may belong to a different model, task or snapshot than the one named beside it, and some sources may come from other snapshots. We have left the post as it was published. Current figures are on the benchmark.

What Changed This Snapshot?

This snapshot does not redraw the model leaderboard. It changes two specific routes: prompt adaptation and language identification.

Meta Muse Spark 1.3 gained 0.12 points on model-specific prompt adaptation, reaching 8.93 out of 10, and 0.06 points on language identification, reaching 9.86.[^3][^1] Those gains are meaningful enough to reconsider the model for these two tasks—but not for translation, voice-profile generation, Markdown repair, or any other unmeasured workload.

The comparison is deliberately like-for-like. Since 2026-09-05, 829 model-task cells remained comparable, while 4 were added and 199 were dropped.[^2] The movement figures therefore describe continuing measurements, not every model and task currently in the benchmark.

There is also a strong reliability signal: meta/muse-spark-1.3-contributor produced a usable first response without repair or retry on 99.9% of 30,332 attempts.[^2] That supports it where first-pass usability matters. It says nothing about whether the model became faster.

The takeaway: update Meta Muse Spark 1.3’s routing for prompt adaptation and language identification. Leave other routes unchanged unless their own task-level evidence moved.

Check the live results at https://fronset.ai/benchmark/ before finalizing a route or comparing batch with synchronous pricing.