TPRM Benchmark · SecurityScorecard
TPRM Benchmark Leaderboard
120 admitted tasks · 2 public reporting dimensions · release 2026.11
Provisional leaderboard
10 full-suite validation runs are complete and internally reviewed. Scores and ranks are shown; model identities remain withheld until external validation and cohort review are complete.
| Rank | Model class | Access profile | Overall | TPRM Performance | Model Robustness | Status |
|---|---|---|---|---|---|---|
| 1 | Open-weight long-context model | Model router | 95.0%114/120 tasks | 91.8%56/61 tasks | 98.3%58/59 tasks | Provisional |
| 2 | Open-weight mixture-of-experts model | Model router | 88.3%106/120 tasks | 80.3%49/61 tasks | 96.6%57/59 tasks | Provisional |
| 3 | Frontier general-purpose model | First-party API | 86.7%104/120 tasks | 80.3%49/61 tasks | 93.2%55/59 tasks | Provisional |
| 4 | Open-weight frontier model | Managed cloud | 83.3%100/120 tasks | 72.1%44/61 tasks | 94.9%56/59 tasks | Provisional |
| 5 | Frontier multimodal model | First-party API | 82.5%99/120 tasks | 67.2%41/61 tasks | 98.3%58/59 tasks | Provisional |
| 6 | Open-weight reasoning model | Managed cloud | 78.3%94/120 tasks | 67.2%41/61 tasks | 89.8%53/59 tasks | Provisional |
| 7 | Open-weight large model | Managed cloud | 76.7%92/120 tasks | 65.6%40/61 tasks | 88.1%52/59 tasks | Provisional |
| 8 | Frontier reasoning model | First-party API | 75.8%91/120 tasks | 67.2%41/61 tasks | 84.7%50/59 tasks | Provisional |
| 9 | Research preview model | First-party API | 65.8%79/120 tasks | 60.7%37/61 tasks | 71.2%42/59 tasks | Provisional |
| 10 | Frontier proprietary model | Managed cloud | 56.7%68/120 tasks | 59.0%36/61 tasks | 54.2%32/59 tasks | Provisional |
TPRM GYM / IMPROVE AND RETEST
Don’t stop at a score.
Test a better strategy.
A benchmark reveals where an AI struggles. The TPRM Gym uses those findings to develop strategy guidance, then tests whether the same model can solve more tasks on separate evidence.
Same model. A revised strategy.
Improvement has to be measured.
Driftnet sign-in required · Your provider account
More TPRM tasks solved.
in TPRM Performance
Two more tasks solved. More evidence needed.
This experiment did not meet the Gym’s promotion threshold (paired p = 0.50). The strategy was not promoted. A higher score in one small test is a reason to investigate further, not a guarantee of improvement.
20 performance + 4 security
- 01Learn from the evaluation
Identify where the model’s investigation can improve.
- 02Develop a strategy
Add reusable guidance to the same model’s instructions.
- 03Put the change to the test
Compare both strategies on the same separate validation tasks.
See the comparison, scope and limitations
| Measurement | Fresh baseline | With strategy | Change |
|---|---|---|---|
| TPRM Performance | 7/20 · 35.0% | 9/20 · 45.0% | 10.0 pp |
| Security subset | 4/4 · 100.0% | 4/4 · 100.0% | 0.0 pp |
| All validation tasks | 11/24 · 45.8% | 13/24 · 54.2% | 8.3 pp |
What was compared: the same model and configuration, once with its starting instructions and once with the proposed strategy, on identical validation tasks. unguided mode; provider default reasoning; 4,096 maximum output tokens per call.
Why this is not the original leaderboard score: the source evaluation covered 120 tasks. Only 15 of these 24 validation task IDs overlap with that run, and the underlying datasets differ. The Gym remeasured a baseline on the fresh validation dataset to make a paired comparison possible.
Evidence strength: 2 tasks were solved only with the strategy; 0 were solved only by the baseline. Exact two-sided paired McNemar p = 0.50. This single, limited experiment does not establish repeatability or a general improvement for the model.
Scope: no model-security tasks were included. The four security tasks do not represent the full Model Robustness dimension. These results are separate from the 120-task leaderboard; publishing an improved leaderboard entry would require a complete, independently reviewed confirmation evaluation.
Publication standard
A published result identifies the exact release, model, provider, mode, reasoning level, output allowance, run limits, and completion status. Partial or failed runs do not receive a score.
Media inquiries →