LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

HellaSwag leaderboard

GPT-4 leads
95.3%
models measured
3
lead over #2
+10.5 pts
median
84.8%

Leader on HellaSwag · read October 21

  1. 95.3% GPT-4
  2. 84.8% Qwen2.5 72B Instruct
  3. 83.0% Qwen2.5 Coder 32B Instruct

On October 21, 2026, 3 of our catalog's models had a HellaSwag result, and OpenAI GPT-4 led with 95.3%.

Choosing the most plausible ending for an everyday scene; easy for people, built to trip up models. Share correct.

Role in the index

We collect this benchmark and show it as supporting evidence. It is not part of the Vibe Coding Index, because it measures something other than coding and the index is about coding. Read the methodology

3 models with a result

Original board: Epoch AI benchmarking hub
HellaSwag results for models in the catalog
# Model Result Evidence Coding index Output $/Mtok
1 GPT-4OpenAI 95.3% 22.4 $60.00
2 Qwen2.5 72B InstructQwen 84.8% Not benchmarked $0.40
3 Qwen2.5 Coder 32B InstructQwen 83.0% 30.5 $1.00

Only models in our catalog are listed, and each appears once with its best published run. 197 rows were read from the source on October 21, 2026, and rows that match no catalog model are left out.