activeMathematics
MATH-500
Metric: Score (higher is better)
500-problem subset of the MATH benchmark covering competition mathematics
Models ranked
7
tracked on this benchmark
Score band
98.0 – 75.2
best → lowest tracked
Snapshot trend
-10.41
Jun 7 → Jun 25 · 1 models
Leaderboard
Tracked models ranked by Score (higher is better).
#Model variant and provenanceRelative to leaderScore
1
Kimi K2.5
98.0Version: Thinking mode (accuracy)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
2
OLMo 3 32B Think
96.1Version: MATH benchmark (accuracy)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
3
Phi-4 Mini Reasoning
94.6Version: From official Microsoft technical report (accuracy)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
4
OLMo 3.1 32B Instruct
93.4Version: MATH benchmark (accuracy)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
5
LFM2.5 8B A1B
88.8Version: MATH-500 (accuracy)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
6
Nemotron-Labs TwoTower 30B-A3B Base
80.6Version: 4-shot; default TwoTower diffusion decoding at confidence_threshold=0.8, block_size=16, BF16 on 2xH100; evaluator/harness not published on the model cardHarness: Not recordedEvaluator: Not recordedObserved: Jun 25, 2026Confidence: Not recordedSource
7
Phi-4 Reasoning Vision 15B
75.2Version: MATH-500 (accuracy)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
How to read this benchmark
This benchmark scores models where higher is better. Use scores for directional filtering and shortlisting, not universal quality ranking; then validate pricing, context window, provider availability, and fit for your workload.
Trust this score when
- There is a fresh timestamped snapshot (or multiple snapshots) for this benchmark.
- The model list covers the same version family you can actually deploy today.
- Top candidates overlap with your required routing and feature requirements.
Be cautious when
- There is only one benchmark snapshot or the dataset appears stale.
- The benchmark metric direction is opposite of your decision objective.
- The score difference between options is narrow and likely within implementation variance.
Related benchmarks
Last reviewed: May 20, 2026