Skip to main content
AI Choice Engine

LG AI Research

K-EXAONE 2.0 (FP8)

K-EXAONE 2.0 (FP8) is a LG AI Research model with 262K context, Not verified in / Not verified out per million tokens, last verified October 2, 2026. Open weights: yes. Public-board scores are linked to the publisher; vendor-only figures stay in the watch-out, not on the board.

LG AI Research’s multilingual open-weight K-EXAONE 2.0, with 262,144-token context and a downloadable FP8 checkpoint.

Recorded profile verification October 2, 2026Individual provider fields may change

AI model specification details
SpecificationK-EXAONE 2.0 (FP8)
ProviderLG AI Research
TierFrontier
Context window262K
Max outputNot verifiedUnverified
Input / 1M tokensNot verifiedUnverified
Output / 1M tokensNot verifiedUnverified
WeightsOpen
Parameters750B total / 37B active (MoE); FP8 checkpoint
Reasoning levelsNot verifiedUnverified
Modalitiestext
LicenseApache 2.0
ReleasedNot verifiedUnverified

Pricing tiers: The exact FP8 checkpoint is downloadable under Apache 2.0; serving infrastructure costs apply. The checked model card links an open demo, but does not establish a first-party hosted API ID or token tariff.

Best for

  • Multilingual text
  • Server-scale self-hosting
  • Long-document work

Watch out

750B total weights still need to be served; 37B active parameters is not a 37B weight footprint. The card's benchmarks describe BF16, so those scores are not assigned to this FP8 profile. enable_thinking defaults true; preserve_thinking defaults false and LG recommends true for long-running agents. No named effort menu or hard output cap is established. July 31 is the family announcement linked to BF16; the exact FP8 checkpoint publication day remains unverified.

Source receipts

Catalog figures for this model were checked against the following sources.

Common questions

K-EXAONE 2.0 (FP8)

Answered from the verified figures on this page rather than general guidance.

What is K-EXAONE 2.0 (FP8)'s context window?
K-EXAONE 2.0 (FP8) accepts about 262K tokens of context. That only matters if you routinely send very long documents, large codebases, or multi-turn histories that approach that limit.
What is K-EXAONE 2.0 (FP8) best for?
K-EXAONE 2.0 (FP8) is a frontier tier from LG AI Research. It suits multilingual text, server-scale self-hosting, long-document work. 750B total weights still need to be served; 37B active parameters is not a 37B weight footprint. The card's benchmarks describe BF16, so those scores are not assigned to this FP8 profile. enable_thinking defaults true; preserve_thinking defaults false and LG recommends true for long-running agents. No named effort menu or hard output cap is established. July 31 is the family announcement linked to BF16; the exact FP8 checkpoint publication day remains unverified.
Can I self-host K-EXAONE 2.0 (FP8)?
K-EXAONE 2.0 (FP8) publishes open weights, but self-hosting depends on the licence, hardware footprint, quantisation quality, and serving stack. A hosted API is often cheaper until you have measured throughput and concurrency on your own hardware.
Model comparison catalogBenchmark library