Rankings / Unknown
Exaone 4.0 32B
released 2024-04-01
BenchAtlas Index
as of 2026-07-14
thinking
rank #354
10 families · 5 categories · high
no reasoning
rank #401
10 families · 5 categories · high
Benchmark evidence
27 results
agentic coding
| τ²-Bench | thinking | 17.3 % | independent | Artificial Analysis |
| τ²-Bench | no reasoning | 4.1 % | independent | Artificial Analysis |
coding
| LiveCodeBench v6 implementation=artificial-analysis | thinking | 74.7 % | independent | Artificial Analysis |
| LiveCodeBench v6 problems=454 · window_start=2024-08-01 | 70.0 % | independent | LiveCodeBench Leaderboard | |
| LiveCodeBench v6 implementation=artificial-analysis | no reasoning | 47.2 % | independent | Artificial Analysis |
| SciCode | thinking | 34.4 % | independent | Artificial Analysis |
| SciCode | no reasoning | 25.2 % | independent | Artificial Analysis |
external indices
| Intelligence Index v4.1 | thinking | 10.6 points | independent | Artificial Analysis |
| Intelligence Index v4.1 | no reasoning | 6.0 points | independent | Artificial Analysis |
| AA Math Index | thinking | 80.0 points | independent | Artificial Analysis |
| AA Math Index | no reasoning | 39.3 points | independent | Artificial Analysis |
knowledge science
| GPQA Diamond implementation=artificial-analysis | thinking | 73.9 % | independent | Artificial Analysis |
| GPQA Diamond implementation=artificial-analysis | no reasoning | 62.8 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | thinking | 10.5 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | no reasoning | 4.9 % | independent | Artificial Analysis |
| MMLU-Pro implementation=artificial-analysis | thinking | 81.8 % | independent | Artificial Analysis |
| MMLU-Pro implementation=artificial-analysis | no reasoning | 76.8 % | independent | Artificial Analysis |
long context instruction
| AA-LCR | thinking | 14.0 % | independent | Artificial Analysis |
| AA-LCR | no reasoning | 8.0 % | independent | Artificial Analysis |
| IFBench | thinking | 36.3 % | independent | Artificial Analysis |
| IFBench | no reasoning | 33.5 % | independent | Artificial Analysis |
reasoning math
| AIME implementation=artificial-analysis | thinking | 84.3 % | independent | Artificial Analysis |
| AIME year=2025 · implementation=artificial-analysis | thinking | 80.0 % | independent | Artificial Analysis |
| AIME implementation=artificial-analysis | no reasoning | 47.0 % | independent | Artificial Analysis |
| AIME year=2025 · implementation=artificial-analysis | no reasoning | 39.3 % | independent | Artificial Analysis |
| MATH-500 implementation=artificial-analysis | thinking | 97.7 % | independent | Artificial Analysis |
| MATH-500 implementation=artificial-analysis | no reasoning | 93.9 % | independent | Artificial Analysis |
Agent + model results
systems, not bare-model scores
| agent + model Artificial Analysis harness + Exaone 4.0 32B | Terminal-Bench Hard | 3.8 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + Exaone 4.0 32B | Terminal-Bench Hard | 1.5 % | independent | Artificial Analysis |
These scores measure the whole agent system (scaffold, tools, budgets) — they are never merged into the bare model’s numbers.