Meta AI·released 2026-08-051 source
Muse Spark 1.2 benchmark scores: 9 benchmarks tracked. 0% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 9 benchmarks·best result 91.12 on DTBench (unverified)·0 of 9 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| DTBench | 91.12accuracy (%) | unverifiedT1 | 2026-08-05https://conceptualreasoning.ai/dtbench |
| SimpleBench | 69.4accuracy (%) | unverifiedT1 | 2026-08-05https://lmcouncil.ai/benchmarks |
| WeirdML | 60.3accuracy (%) | unverifiedT1 | 2026-08-05https://htihle.github.io/weirdml.html |
| SimpleQA Verified | 60.3accuracy (%) | unverifiedT2 | 2026-08-05 |
| LMCA | 56.88accuracy (%) | unverifiedT1 | 2026-08-05https://conceptualreasoning.ai/lmca |
| DeepSWE | 54.87accuracy (%) | unverifiedT1 | 2026-08-05https://deepswe.datacurve.ai/ |
| ProofBench | 43accuracy (%) | unverifiedT1 | 2026-08-05https://www.vals.ai/benchmarks/proof_bench |
| APEX-Agents | 36.4accuracy (%) | unverifiedT2 | 2026-08-05 |
| FrontierSWE | 11.97accuracy (%) | unverifiedT1 | 2026-08-05https://www.frontierswe.com/ |
Claims drawn from cited facts, not live model generation.
This model was developed by Meta AI, placing its origins within one of the major industrial research organizations. It was released in 2026, making it a recent entry rather than a legacy system, though no parameter count is offered here so its scale cannot be stated.
2 cited facts
This model is currently scored on 9 tracked benchmarks, but it holds no top score on any of them. Because no outside evaluation has either confirmed or disputed its numbers, its benchmark record remains unverified.
3 cited facts
Its mean gap to the best recorded score across its tracked benchmarks is 29.35 points.
1 cited fact
1 facts cross-checked across data sources: 1 single-source.
Source facts, citations, and refresh stamp for this record.
Sources: epoch_benchmarks
Muse Spark 1.2 is an AI model developed by Meta AI, released 2026-08-05. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
Muse Spark 1.2 has recorded scores on 9 benchmarks, each shown with its evidence status.
Muse Spark 1.2 has recorded scores on 9 benchmarks — DTBench, WeirdML, LMCA, SimpleQA Verified, SimpleBench, APEX-Agents, and 3 more. The full table above shows each score with its evidence status.
0 of 9 recorded scores (0%) are independently reproduced rather than self-reported by the lab.
Muse Spark 1.2 has 9 tracked claims: 9 unverified.