Skip to main content

FrontierMath v2 Tier 4 (FrontierMath v2 (Tier 4))

Epoch AI's corrected v2 Tier 4 expansion, a separate set of exceptionally difficult research-level mathematics problems evaluated with Python-enabled iterative reasoning.

Data verified

Top models on FrontierMath v2 (Tier 4) — July 9, 2026

As of July 9, 2026, GPT-5.6 Sol leads the FrontierMath v2 (Tier 4) leaderboard with 83.000% , followed by GPT-5.6 Terra (68.300%) and GPT-5.6 Luna (58.500%).

47 modelsMath10% of category scoreCurrentUpdated July 9, 2026

According to BenchLM.ai, GPT-5.6 Sol leads the FrontierMath v2 (Tier 4) benchmark with a score of 83.000%, followed by GPT-5.6 Terra (68.300%) and GPT-5.6 Luna (58.500%). There is significant spread across the leaderboard, making this benchmark effective at differentiating model capabilities.

47 models have been evaluated on FrontierMath v2 (Tier 4). The benchmark falls in the Math category. This category carries a 5% weight in BenchLM.ai's overall scoring system. Within that category, FrontierMath v2 (Tier 4) contributes 10% of the category score, so strong performance here directly affects a model's overall ranking.

About FrontierMath v2 (Tier 4)

Year

2026

Tasks

43 private extreme-difficulty mathematics problems

Format

Python-enabled iterative mathematical problem solving

Difficulty

Research-level mathematics requiring hours or days of expert work

The v2 private Tier 4 set contains 43 problems. Its smaller sample makes individual scores noisier than Tiers 1-3, so it is a separate ranking factor with lower weight rather than being averaged into the core score.

BenchLM freshness & provenance

Version

FrontierMath v2 (Tier 4) 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

Current

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

Leaderboard (47 models)

1
GPT-5.6 Solgpt-5-6-sol
83.000%
2
GPT-5.6 Terragpt-5-6-terra
68.300%
3
GPT-5.6 Lunagpt-5-6-luna
58.500%
4
GPT-5.5 Progpt-5-5-pro
39.600%
5
GPT-5.4 Progpt-5-4-pro
37.500%
6
GPT-5.5gpt-5-5
35.400%
7
GPT-5.2 Progpt-5-2-pro
31.300%
8
Claude Opus 4.8claude-opus-4-8
31.250%
9
GPT-5.4gpt-5-4
27.100%
10
Claude Opus 4.7claude-opus-4-7
22.917%
11
Claude Opus 4.6claude-opus-4-6
22.900%
12
GPT-5.2gpt-5-2
18.800%
13
Gemini 3 Progemini-3-pro
18.750%
14
Gemini 3.1 Progemini-3-1-pro
16.700%
15
Muse Sparkmuse-spark
14.600%
16
Gemini 3.5 Flashgemini-3-5-flash
14.583%
17
Kimi K2.6kimi-2-6
14.580%
18
GPT-5.1gpt-5-1
12.500%
19
GLM-5.1glm-5-1
12.500%
20
Qwen3.6 Plusqwen3-6-plus
8.333%
21
Claude Sonnet 4.6claude-sonnet-4-6
8.300%
22
o4-mini (high)o4-mini-high
6.250%
23
GPT-5.4 nanogpt-5-4-nano
6.250%
24
GPT-5 minigpt-5-mini
6.250%
25
Kimi K2.5kimi-k2-5
4.200%
26
Claude Opus 4.5claude-opus-4-5
4.167%
27
Claude Sonnet 4.5claude-sonnet-4-5
4.167%
28
Gemini 2.5 Progemini-2-5-pro
4.167%
29
Gemini 3 Flashgemini-3-flash
4.167%
30
Gemini 2.5 Flashgemini-2-5-flash
4.167%
31
Qwen 3.6 Max (preview)qwen3-6-max-preview
4.167%
32
GLM-4.6glm-4-6
2.128%
33
GLM-5glm-5
2.100%
34
DeepSeek V3.2deepseek-v3-2
2.100%
35
Grok 4grok-4
2.083%
36
2.083%
37
Claude Haiku 4.5claude-haiku-4-5
2.083%
38
Qwen3.5 Plusqwen3-5-plus
2.083%
39
GPT-5.4 minigpt-5-4-mini
2.080%
40
GLM-4.7glm-4-7
0.000%
41
GPT-4.1gpt-4-1
0.000%
42
Qwen3 235B 2507 (Reasoning)qwen3-235b-2507-reasoning
0.000%
43
Kimi K2kimi-k2
0.000%
44
Claude 3.5 Sonnetclaude-3-5-sonnet
0.000%
45
Grok 3 [Beta]grok-3-beta
0.000%
46
Qwen3.5 Flashqwen3-5-flash
0.000%
47
GPT-5 nanogpt-5-nano
0.000%

FAQ

What does FrontierMath v2 (Tier 4) measure?

Epoch AI's corrected v2 Tier 4 expansion, a separate set of exceptionally difficult research-level mathematics problems evaluated with Python-enabled iterative reasoning.

Which model scores highest on FrontierMath v2 (Tier 4)?

GPT-5.6 Sol by OpenAI currently leads with a score of 83.000% on FrontierMath v2 (Tier 4).

How many models are evaluated on FrontierMath v2 (Tier 4)?

47 AI models have been evaluated on FrontierMath v2 (Tier 4) on BenchLM.

Last updated: July 9, 2026 · BenchLM version FrontierMath v2 (Tier 4) 2026

The AI models change fast. We track them for you.

For engineers, researchers, and the plain curious — a weekly brief on new models, ranking shifts, and pricing changes.

Free. No spam. Unsubscribe anytime.