Clinical Benchmarks

Meta

Meta models with results here, each shown on its own board.

Models
9
Results
22
Boards
8 of 14

ai.meta.com

Models

ModelKindReleasedBoards
Muse Spark OpenModel15 Oct 20262
Muse WonderModel12 Oct 20263
Muse Wonder FullModel12 Oct 20262
Muse Spark 1.3Model2 Sep 20261
Muse Spark 1.2Model5 Aug 20263
Muse Spark 1.1Model9 Jul 20263
Muse SparkModel8 Apr 20264
Llama 4 MaverickModel5 Apr 20252
Llama 4 ScoutModel5 Apr 20252

Results

One line per result. Tall tick: the board leader.

Clinical reasoning and knowledge

  1. HealthBench Professional
    Meta internal run; length-normalized weighted rubric score, GPT-5.4 grader, no tools, max reasoning effort (Muse Wonder Model Card p. 3 table, Muse Wonder Full (max) column)
    0.623
    Rank 11 of 43Leader GPT-6 Astra (Anthropic run) 0.703
    Vendor-reported
    Measured Oct 2026
  2. HealthBench Professional
    Meta internal run; length-normalized weighted rubric score, GPT-5.4 grader, no tools, max reasoning effort (Muse Wonder Model Card p. 3 table, Muse Wonder (max) column)
    0.605
    Rank 15 of 43Leader GPT-6 Astra (Anthropic run) 0.703
    Vendor-reported
    Measured Oct 2026
  3. HealthBench Professional
    length-normalized, GPT-5.4 low-reasoning grader, xhigh reasoning via Meta Model API (Muse Spark 1.1 Evaluation Report Figure 44)
    0.593
    Rank 18 of 43Leader GPT-6 Astra (Anthropic run) 0.703
    Vendor-reported
    Measured Jul 2026
  4. HealthBench Professional
    length-normalized, GPT-5.4 low-reasoning grader; Muse Spark (1.0) column in the Muse Spark 1.1 Evaluation Report Figure 44
    0.541
    Rank 26 of 43Leader GPT-6 Astra (Anthropic run) 0.703
    Vendor-reported
    Measured Jul 2026
  5. MedXpertQA (MM)
    Meta's Muse Spark launch table (Meta reports the better of vendor self-reports and its own reproduction)
    78.4
    Rank 5 of 22 here, 5 models on the boardLeader GPT-5.6 Sol 81.5
    Vendor-reported
    Measured Apr 2026
  6. 94.3
    Rank 2 of 15 here, 16 models on the boardLeader Claude Opus 5 94.6
    Official leaderboard
    Measured Oct 2026

Documentation and coding

  1. 90.06
    Rank 5 of 106 here, 105 models on the boardLeader Claude Opus 5.5 91.43
    Official leaderboard
    Measured Sep 2026
  2. 88.89
    Rank 8 of 106 here, 105 models on the boardLeader Claude Opus 5.5 91.43
    Official leaderboard
    Measured Sep 2026
  3. MedScribe (Vals AI)
    model ID meta/muse_spark; temperature=1; max_output_tokens=30000
    85.90
    Rank 23 of 106 here, 105 models on the boardLeader Claude Opus 5.5 91.43
    Official leaderboard
    Measured Sep 2026
  4. MedScribe (Vals AI)
    model ID fireworks/llama4-maverick-instruct-basic; temperature=1; max_output_tokens=30000
    54.22
    Rank 104 of 106 here, 105 models on the boardLeader Claude Opus 5.5 91.43
    Official leaderboard
    Measured Sep 2026
  5. MedScribe (Vals AI)
    model ID together/meta-llama/Llama-4-Scout-17B-16E-Instruct; temperature=1; max_output_tokens=30000
    50.59
    Rank 105 of 106 here, 105 models on the boardLeader Claude Opus 5.5 91.43
    Official leaderboard
    Measured Sep 2026
  6. MedCode (Vals AI)
    model ID meta/muse_spark; temperature=1; max_output_tokens=30000
    51.31
    Rank 15 of 104 here, 103 models on the boardLeader Claude Opus 5 63.57
    Official leaderboard
    Measured Sep 2026
  7. MedCode (Vals AI)
    model ID meta/muse_spark_1_2; reasoning_effort=xhigh; temperature=1; max_output_tokens=30000
    49.35
    Rank 21 of 104 here, 103 models on the boardLeader Claude Opus 5 63.57
    Official leaderboard
    Measured Sep 2026
  8. MedCode (Vals AI)
    model ID fireworks/llama4-maverick-instruct-basic; temperature=1; max_output_tokens=30000
    36.51
    Rank 75 of 104 here, 103 models on the boardLeader Claude Opus 5 63.57
    Official leaderboard
    Measured Sep 2026
  9. MedCode (Vals AI)
    model ID together/meta-llama/Llama-4-Scout-17B-16E-Instruct; temperature=1; max_output_tokens=30000
    23.31
    Rank 101 of 104 here, 103 models on the boardLeader Claude Opus 5 63.57
    Official leaderboard
    Measured Sep 2026
  10. MLCR-AA
    overall pass rate, Max
    22.2
    Rank 13 of 27Leader Claude Sonnet 5.5 (Max, Default Fallback) 41.7
    Independent run
    Measured Nov 2026
  11. MLCR-AA
    overall pass rate, Max
    22.2
    Rank 13 of 27Leader Claude Sonnet 5.5 (Max, Default Fallback) 41.7
    Independent run
    Measured Nov 2026
  12. MLCR-AA
    overall pass rate, Xhigh
    7.8
    Rank 26 of 27Leader Claude Sonnet 5.5 (Max, Default Fallback) 41.7
    Independent run
    Measured Nov 2026

EHR and workflow agents

  1. DAYJOB Health
    Score (pass@1 at 95%), reasoning max
    31.2
    Rank 3 of 15Leader Claude Opus 5.5 48.3
    Official leaderboard
    Measured Oct 2026
  2. DAYJOB Health
    Score (pass@1 at 95%), reasoning max
    22
    Rank 9 of 15Leader Claude Opus 5.5 48.3
    Official leaderboard
    Measured Oct 2026
  3. DAYJOB Health
    Score (pass@1 at 95%), reasoning xhigh
    8.8
    Rank 13 of 15Leader Claude Opus 5.5 48.3
    Official leaderboard
    Measured Oct 2026

Safety

  1. 79.7
    Rank 4 of 17 here, 19 models on the boardLeader LiSA 2.5 86.2
    Official leaderboard
    Measured Aug 2026