Model

ARPO

An 8B agent model with a pure-text, no-latent-space mode, serving as a comparison baseline for HybridAgent.

Timeline (last 90 days)

Oct 19

Table 1 of the paper shows that ARPO (a pure-text 8B agent without latent-space mode) scores 38.8 on GAIA and 32.0 on WebWalker, both slightly higher than GPT-5.6 Luna.

Unverified50%

Oct 19

The incremental gains from latent-space mode are: GAIA rises from 38.8 to 43.7, and WebWalker jumps from 32.0 to 42.5.

Unverified50%

All Facts (2)

Unverified

Table 1 of the paper shows that ARPO (a pure-text 8B agent without latent-space mode) scores 38.8 on GAIA and 32.0 on WebWalker, both slightly higher than GPT-5.6 Luna.

50%Unverified

The incremental gains from latent-space mode are: GAIA rises from 38.8 to 43.7, and WebWalker jumps from 32.0 to 42.5.

50%

Source Articles

8B Agent Beats GPT-5.6 Luna on GAIA with Latent-Space Reasoning, but Crumbles on HLEOct 19