Model

Llama-3.1-ELYZA-JP-70B

CLOSED

ELYZA · ELYZA LLM for JP family · released Oct 25, 2024

Latest ELYZA LLM for JP flagship, built on Llama 3.1 for longer context and improved Japanese instruction-following vs. the earlier Llama-3-based 70B; distributed only via demo and enterprise API/joint-development deals, not as public open weights.

ReasoningCodingVisionFunction callingTool useAgentic

1190.8

Elo · unrated

Parameters

70B

Active params

70B (dense)

Context

128K tokens

Architecture

70B dense transformer (Llama-3.1 architecture, Japanese continual pretrain + instruction tuning)

License

Proprietary — enterprise/partner access only (no public license terms published; not an open-weight release)

Languages

API price (in/out)

No hosted API

Modalities

text

Benchmark results

Bar shows position within the tracked field; marker = field best

MT-BenchHuman preference9.08#8

best: Hunyuan-Large (A52B) · 9.4

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

—

Output / M tok

—

ELYZA LLM for JP family

Elo progression across releases

Llama-3-ELYZA-JP-8BJun 2024324.8 Llama-3.1-ELYZA-JP-70BOct 20241190.8

Resources

DOCSelyza.ai

API price — · each benchmark row carries its own source badge (see methodology)