SEA-LION v3.5 (Latest)

SEA-LION version 3.5, released in Apr 2025, is our first set of hybrid reasoning models trained on Southeast Asian data, each with their unique strengths:

• Llama-SEA-LION-v3.5-8B-R is suited for cost-efficient, low-latency applications where regional language support and advanced reasoning is critical.

• Llama-SEA-LION-v3.5-70B-R is suited for knowledge-intensive tasks, and for high-demand environments where advanced reasoning and comprehensive language comprehension are essential.

Trained off our SEA-LION v3 models, SEA-LION v3.5 is explicitly enhanced for reasoning tasks with the inclusion of thinking blocks, continuing our mission to create language models that understand and respond with greater cultural awareness and depth across Southeast Asia.

A distinctive feature of SEA-LION v3.5 is its dynamic reasoning toggle - Mode selection is managed through customizable chat template configurations, offering versatile functionality in handling both complex reasoning tasks and general text generation.

For detailed information of each of the SEA-LION v3.5 models, please refer to their individual documentation pages via the links above.