We're excited to announce that our latest language model, Spicy 2.1, has been promoted to become the new default model for all users. Based on the feedback received after our previous model update in May, we have focused on enhancing the response quality while maintaining a desirable output length.

Key Improvements in Spicy 2.1

• Longer Responses: The new model generates responses that are on average 50% longer than our previous default, providing more comprehensive and detailed outputs.
• Quality Boost: We have measured an 8% reduction in negative ratings, indicating a significant improvement in the overall quality and accuracy of the responses.

Upcoming Models for I'm All In Tier

For our valued I'm All In tier members, we have two exciting new models in the pipeline:

1. Command-R 34B

Renowned for its creativity and exceptional prompt-following capabilities, the highly anticipated Command-R 34B model is scheduled to be added around June 26th.

2. WizardLM-2 8x22B

Prepare to be amazed by the sheer power of WizardLM-2 8x22B, a massive 141B parameter model. This cutting-edge model will be available for a limited 30-day period starting around July 3rd, allowing us to assess its performance and cost-effectiveness before deciding on its long-term availability. We are committed to continuously improving our language models and providing our users with the best possible experience. Stay tuned for more updates and exciting developments in the near future.