Back to All Resources

Qwen2.5-14B-Instruct-1M

Scales up Qwen's 1M-token context capabilities to 14B parameters. Maintains high throughput with optimized attention mechanisms and improved memory management. Released alongside comprehensive technical documentation.