A text-focused model with an unusually large context window of 262,144 tokens, allowing it to process and reason over very long documents in a single pass. Released as open weights in safetensors format, it can be run and modified locally. Specific capability details beyond text-in, text-out processing are limited from available information.