SectorFlow v0.5 Release Notes

Unveiling SectorFlow v0.5, the latest and most substantial update to our platform yet. This version marks a significant milestone in the evolution of SectorFlow, introducing a vast array of new AI models, a complete overhaul of our communication systems, enhanced team functionalities, and substantial performance improvements.

What's New:

Expansion of AI Model Portfolio

With v0.5, SectorFlow broadens its horizons by integrating 8 new AI models, offering diversity and power in language processing capabilities:

New AI Models: Unlocking Next-Level Capabilities

Amazon Titan Text G1 (8K tokens): This large language model from Amazon excels at text generation and conversational chat. It supports English and preview support for 100 other languages.
Claude 3 Opus (200,000 token context window): Anthropic's groundbreaking model, pushing the boundaries of language model performance with unparalleled context understanding, especially those involving coding capabilities.
Cohere Command R (128,000 token context window): Cohere's optimized model for conversational interactions, fluent in 10 languages with support for 13 additional languages.
Meta Llama 3 (8B and 70B): Experience the state-of-the-art capabilities of Meta's Llama 3 models, now available in 8B and 70B versions. With improved pretraining, post-training procedures, and a 8,192 token context window, these models excel in reasoning, code generation, and instruction following.
Microsoft Phi 3 Mini (3.8B parameters): Microsoft's Small Language Model (SLM) is compact yet powerful model designed for simpler tasks, offering better performance than models twice its size. Ideal for summarizing documents, extracting insights, generating content, and powering chatbots.
Mixtral 8x22B Instruct (39B active parameters): A cost-efficient, sparse Mixture-of-Experts model with strong mathematics, coding, and function-calling capabilities, supporting 5 languages and a 64K token context window.
Snowflake Arctic 17B (4,096 token context window): Boasting enhanced context understanding and generation, Arctic excels at enterprise tasks like SQL generation, coding, and instruction following, setting new benchmarks for cost-effective training.

Sunsetting AI Models

As part of our commitment to providing the best and most up-to-date AI capabilities, we are sunsetting the following models:

Claude Instant
Claude 2.1
LLama 2 13b Chat
LLama 2 70b Chat
OpenAI ChatGPT 4

These models are being retired based on a combination of factors, including their placement on the LMSYS Chatbot Arena Leaderboard, our own internal tests, and comparisons with later versions within the same model families. As new and more capable models are released, we continually evaluate and update our AI model portfolio to ensure we are offering the best possible experience to our users.

Communication Overhaul

We’ve redesigned all outbound emails from SectorFlow to enhance clarity, functionality, and user engagement, ensuring you’re well-informed and connected at every touchpoint.

Enhanced Team Features

Credit Balances: Each team will now see their credit balances directly, fostering better resource planning.
API Keys: New functionality allows teams to create secure tokens, granting safe and controlled API access.
Billing and Team Settings: Updated sections for better management, including tabs to control AI model access, options to manage projects individually or in bulk, and visual representations of credit usage.
Credits Dashboard: Improved tracking and visualization of credit expenditure and allocations to maximize efficiency.

Performance Improvements

Expect smoother operation and faster responses across the board thanks to major backend optimizations designed to enhance your SectorFlow experience.

We are committed to continuously refining SectorFlow to meet and exceed the needs of our diverse user base. Thank you for your continued trust and support. Enjoy the new features, and we look forward to your invaluable feedback.

– The SectorFlow Team