Reka Core

Multimodal language model developed by Reka AI for agentic reasoning across vision, audio, and text-based tasks:

• Processes and integrates inputs from text, images, video, and audio using a unified 67B-parameter architecture
• Features an extended 128K token context window and native speech output for conversational agent use
• Outperforms or rivals GPT-4V, Claude 3 Opus, and Gemini Ultra on reasoning and multimodal benchmarks
• Designed by Reka AI, a research-driven company founded by former DeepMind scientists
• Available through Reka’s hosted chat interface and API under a proprietary cloud-access license

CLICK HERE TO DISCOVER REKA CORE