The development
What happened
Swiss AI’s 15 October report explains how Apertus 1.5 adds image and experimental audio input to its 8B and 70B models. The weights were released in July; the new coverage concerns technical documentation, training materials and evaluation transparency.
Key facts
Both sizes generate text, with optional thinking and a context window of 262,144 tokens.
Apache 2.0 weights are accompanied by training-data documentation, recipes and intermediate checkpoints.
Evaluation records use the Every Eval Ever format to identify who ran a test, generation settings and metric context.
The release notes discuss models available since 24 July, rather than announcing a new October weight release.
The context
Why it matters
TideReach assessment: Published intermediate stages and evaluation settings make it easier to inspect which training choices changed model behavior. Availability of those materials supports reproducibility, but does not independently validate the developers’ performance comparisons.