Open Source

Mistral Open Source

Mistral Small 3.1 Outperforms Gemma 3 and GPT-4o Mini

• 

Chris McKay• March 17, 2025 • 2 min read

Mistral AI has released Mistral Small 3.1, a 24-billion-parameter open-source model that outperforms Google’s Gemma 3 and OpenAI’s GPT-4o Mini in key benchmarks, offering improved multimodal capabilities, long-context processing, and affordability for local deployment.

Google Open Source

Google DeepMind’s Gemma 3 Brings Multimodal AI, 128K Context Window, and More

• 

Chris McKay• March 12, 2025 • 2 min read

Google DeepMind has launched Gemma 3, the latest iteration of its open AI model family, introducing multimodal capabilities, a 128K-token context window, and a range of model sizes optimized for various computing environments.

Open Source

SambaNova and CrewAI Launch Open-Source Deep Research Framework for Faster, Cheaper AI-Powered Insights

• 

Chris McKay• March 10, 2025 • 3 min read

SambaNova and CrewAI have introduced an open-source Deep Research framework that accelerates enterprise AI research by 3X, slashes costs by millions annually, and ensures data security through on-premise deployment.

DeepSeek China Open Source

DeepSeek R1 is a Fully Open-Source "Reasoning" AI Model

• 

Chris McKay• January 20, 2025 • 2 min read

DeepSeek-R1 is not just a model but a statement on behalf of the open source community. It offers performance that matches OpenAI's o1 in critical areas such as reasoning, mathematics, and coding.

China Open Source

MiniMax Releases Open-Source Model with Massive 4M Context Window

• 

Chris McKay• January 14, 2025 • 2 min read

It is 10x cheaper than GPT-4o, with API costs at $0.20/M input tokens and $1.10/M output tokens.

DeepSeek China Open Source

DeepSeek-V3 Achieves Frontier AI Performance at a Fraction of the Cost

• 

Chris McKay• December 27, 2024 • 4 min read

DeepSeek's new AI model demonstrates that top-tier performance doesn't require massive resources, raising questions about the economics of AI development.

Alibaba China Open Source

Alibaba Unveils QVQ, an Open-Weight AI Model Focused on Visual Reasoning

• 

Chris McKay• December 24, 2024 • 2 min read

QVQ is built upon Qwen2-VL-72B's architecture but incorporates refinements tailored to multimodal reasoning.

UAE Open Source

UAE's TII Launches Falcon 3: High-Performance Small AI Models

• 

Chris McKay• December 17, 2024 • 2 min read

Trained on 14 trillion tokens—over double its predecessor Falcon 2—Falcon 3 achieves superior performance in benchmarks, topping Hugging Face’s leaderboard for models under 13 billion parameters.

Open Source

Harvard Releases Massive Public-Domain Dataset for AI Training

• 

Chris McKay• December 12, 2024 • 2 min read

With initial funding from Microsoft and OpenAI, the project is focused on refining and releasing data collections that can support AI development while honoring principles of accessibility and stewardship.

Meta Open Source

Meta Releases Video Seal, an Open Source Watermarking Tool for AI Video

• 

Chris McKay• December 12, 2024 • 2 min read

As AI-generated content becomes increasingly sophisticated and prevalent, robust watermarking solutions like Video Seal will play a crucial role in content verification and digital rights management.

Meta Llama Open Source

Meta Wraps Up 2024 with the Release of Llama 3.3

• 

Chris McKay• December 6, 2024 • 2 min read

This new model is more efficient, cost-effective, and accessible for developers seeking high-quality text-based AI solutions.

Open Source Media & Entertainment

Tencent Launches HunyuanVideo, an Open-Source AI Video Model

• 

Chris McKay• December 3, 2024 • 2 min read

The 13-billion parameter open-source model doesn't just generate videos - it can add synchronized sound effects, animate avatars, and transform still images into cinematic sequences.

AI2 Open Source

Ai2 Releases OLMo 2,the Most Capable Fully Open AI Model

• 

Chris McKay• November 27, 2024 • 1 min read

OLMo 2 outperforms other fully open models and competes with open-weight models like Llama 3.1 8B,

Anthropic Open Source

Anthropic Launches Open Protocol to Connect AI with Enterprise Data

• 

Chris McKay• November 25, 2024 • 2 min read

Developers can use it to build secure, two-way connections between AI tools and diverse data sources.

DeepSeek China Open Source

Chinese AI Lab DeepSeek Challenges OpenAI with New Reasoning Model

• 

Chris McKay• November 20, 2024 • 2 min read

You can try the model via their web-based chatbot, DeepSeek Chat and assess it for yourself.

China Open Source

Tencent Unveils Hunyuan-Large, an Open-Source MoE Model that Outperforms Llama

• 

Chris McKay• November 5, 2024 • 2 min read

This is the largest open-source Transformer-based MoE model to date, capable of handling sequences up to 256K tokens.

Cohere Open Source

Cohere for AI Launches Aya Expanse to Help Close the Language Gap

• 

Chris McKay• October 25, 2024 • 2 min read

The models can understand and process 23 languages more effectively than similar tools from Google, Meta, and Mistral – despite often using fewer computing resources.

IBM Open Source

IBM and NASA Release Prithvi, an Open-Source Foundation Model for Climate and Weather

• 

Chris McKay• September 23, 2024 • 2 min read

Prithvi can be fine-tuned for various tasks with minimal additional data, demonstrating remarkable efficiency. In one experiment, it accurately reconstructed global surface temperatures using only 5% of the original data.

Google Open Source

Google Introduces DataGemma to Improve AI Accuracy with Trusted Data Sources

• 

Chris McKay• September 12, 2024 • 3 min read

The models are grounded in Google’s Data Commons—a repository of over 240 billion data points collected from trusted organizations like the United Nations and the Centers for Disease Control and Prevention.

Biotech & Health Open Source

Chai Discovery Releases Powerful New Open AI Model for Molecular Structure Prediction

• 

Chris McKay• September 10, 2024 • 2 min read

Chai-1 enables unified prediction of proteins, small molecules, DNA, RNA, and other biomolecular structures. It matches or exceeds the performance of leading models like AlphaFold 3 on several benchmarks.

EU Open Source

Aleph Alpha Releases Pharia: Open-Weight, EU-Compliant, AI Models

• 

Chris McKay• August 27, 2024 • 3 min read

The company claims that these models comply with the General Data Protection Regulation (GDPR) and are designed to meet the forthcoming obligations of the EU AI Act.

AI21 Open Source

AI21 Releases Jamba 1.5: A Family of Open Models with Long Context and Low Latency

• 

Chris McKay• August 22, 2024 • 2 min read

At the heart of these models is a hybrid SSM-Transformer architecture that combines Transformer and Mamba layers with a Mixture-of-Experts approach. This innovative design allows Jamba models to process vast amounts of information with exceptional speed and accuracy.

Open Source

TII Unveils Falcon Mamba 7B, A New Open-Source State Space Language Model

• 

Chris McKay• August 12, 2024 • 2 min read

These models can handle longer sequences of text more efficiently, require less memory to operate, and maintain consistent performance regardless of input size.

Open Source

LG Releases EXAONE 3.0: First Open-Weights Korean Language Model

• 

Chris McKay• August 7, 2024 • 2 min read

It has the potential to accelerate AI research and development in Korea and beyond, possibly leading to breakthroughs in natural language processing for the Korean language.

Google Open Source

Google Expands Gemma 2 Family with New 2B Model, Safety Classifiers, and Interpretability Tools

• 

Chris McKay• July 31, 2024 • 2 min read

Google has unveiled a trio of AI advancements, expanding its Gemma 2 family with a potent small-scale model and introducing new tools for safety and transparency. The star of the show is Gemma 2 2B, a new 2 billion parameter model that punches well above its weight class. Despite its