activeMathematicsReasoning

FrontierMath Tier 4

Metric: Accuracy (higher is better)Introduced: 2026

FrontierMath Tier 4 research-mathematics benchmark tier for frontier reasoning models.

Models ranked

tracked on this benchmark

Score band

100.0 – 0.0

best → lowest tracked

Snapshot trend

-23.09

Sep 22 → Oct 17 · 3 models

Leaderboard

Tracked models ranked by Accuracy (higher is better).

Compare candidates

#Model variant and provenanceRelative to leaderScore

GPT-6.1 Sol

Configuration: gpt-6.1-sol (Max)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Sep 29, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'gpt-6.1-sol_max', run started 2026-09-29). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

100.0

GPT-6 Astra

Configuration: gpt-6-astra (Medium)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Aug 30, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'gpt-6-astra_medium', run started 2026-08-30). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

97.6

Claude Opus 5.5

Configuration: claude-opus-5-5 (Max)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Sep 22, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'claude-opus-5-5_max', run started 2026-09-22). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

95.0

GPT-6 Sol

Configuration: gpt-6-sol (Max)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Sep 22, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'gpt-6-sol_max', run started 2026-09-22). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

90.0

Claude Fable 5.1

Configuration: claude-fable-5-1 (Max)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Sep 1, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'claude-fable-5-1_max', run started 2026-09-01). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

87.8

Claude Sonnet 5.5

Configuration: claude-sonnet-5-5 (Max)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Sep 29, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'claude-sonnet-5-5_max', run started 2026-09-29). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

80.5

Gemini 4 Argon

Configuration: gemini-4-argon (High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Oct 17, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4.csv, row 'gemini-4-argon_high', run started 2026-10-13). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities Benchmarking Hub'. Retrieved 2026-10-17.

78.0

GPT-6 Luna

Configuration: gpt-6-luna (Max)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Sep 22, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'gpt-6-luna_max', run started 2026-09-22). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

56.1

Muse Spark 1.3

Configuration: muse-spark-1.3 (Max)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Sep 18, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'muse-spark-1.3_max', run started 2026-09-18). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

46.3

Qwen3.8-Max

Configuration: qwen3.8-max (Extra High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Aug 4, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'qwen3.8-max_xhigh', run started 2026-08-04). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

46.3

GPT-5.5 Pro

Version: FrontierMath Tier 4; GPT-5.5 Pro computeHarness: Not recordedEvaluator: Not recordedObserved: Apr 24, 2026Confidence: Not recordedSource

Notes: One of the explicitly Pro-differentiated scores. No Claude Opus 4.8 FrontierMath Tier 4 score was found, so this is GPT-only evidence.

39.6

Kimi K3

Configuration: kimi-k3 (Max)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Jul 17, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'kimi-k3_max', run started 2026-07-17). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

39.0

Gemini 3.7 Flash

Configuration: gemini-3.7-flash (High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Aug 14, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'gemini-3.7-flash_high', run started 2026-08-14). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

36.6

Qwen3.8-Max-0902

Configuration: qwen3.8-max-0902 (Extra High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Sep 2, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'qwen3.8-max-0902_xhigh', run started 2026-09-02). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

34.1

Grok 4.6

Configuration: grok-4.6 (Extra High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Aug 14, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'grok-4.6_xhigh', run started 2026-08-14). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

31.7

GLM-5.3

Configuration: glm-5.3 (Max)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Aug 25, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'glm-5.3_max', run started 2026-08-25). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

29.3

Mistral Large 4

Configuration: mistral-large-4 (High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Oct 17, 2026Confidence: confirmedSource

26.8

Gemini 3.6 Flash

Configuration: gemini-3.6-flash (High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Aug 2, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'gemini-3.6-flash_high', run started 2026-08-02). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

22.0

Gemini 3.8 Flash

Configuration: gemini-3.8-flash (High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Sep 2, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'gemini-3.8-flash_high', run started 2026-09-02). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

22.0

GPT-5 Pro

Configuration: gpt-5-pro-2025-10-06 (High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Oct 17, 2026Confidence: confirmedSource

19.5

Gemini 3 Flash Preview

Configuration: gemini-3-flash-preview (single reported configuration)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Jun 11, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'gemini-3-flash-preview', run started 2026-06-11). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

17.1

GLM-5.3-Flash

Configuration: glm-5.3-flash (Max)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Aug 27, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'glm-5.3-flash_max', run started 2026-08-27). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

17.1

Grok 4.7

Configuration: grok-4.7 (Extra High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Sep 22, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'grok-4.7_xhigh', run started 2026-09-22). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

17.1

Inkling-Small

Configuration: Inkling-Small (Extra High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Aug 14, 2026Confidence: confirmedSource

Notes: Epoch AI-run; frontiermath_tier_4_v2.csv row 'inkling-small_xhigh' 0.17073 (run 2026-08-14). CC BY 4.0.

17.1

Inkling

Configuration: Inkling (Extra High)

Version: FrontierMath Tier 4 v2 (private)Harness: Epoch AI FrontierMath Tier 4 v2 private setEvaluator: Epoch AIObserved: Aug 6, 2026Confidence: confirmedSource

Notes: Epoch AI-run evaluation (evaluator of record), from Epoch AI Benchmarking Hub data download https://epoch.ai/data/benchmark_data.zip (file frontiermath_tier_4_v2.csv, row 'Inkling_xhigh', run started 2026-08-06). Epoch data is CC BY 4.0: credit 'Epoch AI, Capabilities & benchmarking, https://epoch.ai/benchmarks'. Highest-scoring Epoch configuration for this model selected; other configurations exist in the same file. Discovered via Benchmark Heaven (lead only).

4.9

Showing the top 25 of 26 tracked models. Browse all models.

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

Average Normalized Levenshtein Similarity*

ARC

AI2 Reasoning Challenge

ARC-AGI-2

Last reviewed: Jun 17, 2026

Resources

Website