EKA-1

Announced

Soket AI LabsOpen source

26.3B-parameter open sparse hybrid foundation model trained on ~7.5T tokens across all 22 scheduled Indian languages plus English, for math, code, reasoning and agentic use (IndiaAI Mission). Early access via waitlist.

Official page ↗

Specifications

License: Open source
Weights: Not released
Architecture: hybrid
Parameters: 26.3B
Context window: — tokens
Max output: —
Knowledge cutoff: —
Price (in / out, $/M): —
Modalities: TextCode

Benchmarks

No benchmark scores recorded yet. Spotted some? Submit a correction.

Vendor-reported figures are claims until independently verified. See methodology.