Overview

EXAONE Deep adds reasoning training to the EXAONE 3.5 models; LG reported the 32B version competing with much larger reasoning models on math benchmarks.

Highlights

  • 2.4B, 7.8B and 32B sizes
  • Chain-of-thought reasoning
  • Math and coding focus
  • Same architecture as EXAONE 3.5

Typical uses

Local reasoning experiments and research.

Things to consider

  • Non-commercial license.
  • Long outputs.
  • Download sizes refer to the default quantized versions in the Ollama library; full-precision weights are larger.

Specifications are taken from the model card on Hugging Face, the Ollama library and the provider’s announcement. Benchmark results depend on evaluation settings; see how AI benchmarks work.