d1-omni-600M

We track d1-omni-600M, but the public leaderboard excludes this profile until enough non-generated benchmark coverage is available. Only published rows appear above.

d1-omni-600M is a open weight model with a 16384 tokens context window. No explicit reasoning mode is documented in this profile.

This experimental checkpoint has 587M parameters and accepts text with images or text with audio, never images and audio in the same request. Its serving config caps the combined context at 16,384 positions and image-request text at 896 tokens, despite the trunk config listing 128,000 positions. Speech clips are capped at 30 seconds; the audio training covers English assistant requests. Liquid reports 78.4% mean accuracy on seven text decision tasks, omits HelpSteer2 because of possible training overlap, and publishes no inference timings for this checkpoint. The weights license has commercial-use conditions.

Liquid AI released d1-omni-600M on October 7, 2026, under the LFM Open License v1.0. It returns Noul, Choice, or Score answers in one forward pass with zero generated tokens. This experimental checkpoint has 587M parameters and accepts text with images or text with audio, never images and audio in the same request. Its serving config caps the combined context at 16,384 positions and image-request text at 896 tokens, despite the trunk config listing 128,000 positions. Speech clips are capped at 30 seconds; the audio training covers English assistant requests. Liquid reports 78.4% mean accuracy on seven text decision tasks, omits HelpSteer2 because of possible training overlap, and publishes no inference timings for this checkpoint. Liquid AI evaluated these checkpoints with the official Decision Index 0.2.1 scorer on the public split. These are provider-reported runs, not public leaderboard submissions. The index and five areas are skill points, not percent accuracy. They remain separate from Decision Index 0.3, JevBench, and general model rankings; no private vision-split result is reported.

d1-omni-600M sits in the Liquid d1 open family with d1-3B. 6 of 665 tracked benchmark slots currently have displayable evidence. Missing categories stay blank.