Llama
Stanford's instruction-tuned Llama model — trained on 52k GPT-generated instructions for $600.
Top-ranked instruction model from StabilityAI — trained with System Prompt Tuning for superior instruction following.
Meta's open-source code LLM family — specialised versions for Python, instruction following, and code infill.
QLoRA-fine-tuned open-source chatbot that achieves ChatGPT quality using only 24GB GPU memory.
Berkeley's open-source chatbot trained on dialogue data from real users — excellent conversational naturalness.
Community fine-tuned instruction model from Nous Research — top performer on open-source model leaderboards.
Microsoft's small language model trained to reason like GPT-4 by learning from complex explanations.
Open-source LLM fine-tuned on STEM and logic questions — top benchmark scorer with minimal training data.
Open-weight fine-tunes from Abacus AI
Fine-tuned open-source chatbot trained on ShareGPT conversations — 90% quality of ChatGPT at minimal cost.
Microsoft's instruction-following LLM trained with Evol-Instruct — progressively harder instructions produce top results.
Llama 4 Scout and Maverick Close the Gap on Frontier Models — With a 10-Million-Token Context Window
Meta's Llama 4 Scout ships a 10-million-token context window — 10× anything in the open-source field. Maverick...
AI
10 May
Open Source AI Is Closing the Gap — A 7B Model Now Does What a 70B Model Did One Year Ago
The latest AI trends data shows open-weight models from Meta, Mistral, and Alibaba's Qwen are matching or exce...
RE
1 May
Meta AI Reaches 1 Billion Users — But Earns Just 10 Cents Per Head
Meta AI reached 1 billion monthly active users but earns just $0.10 per user. Here is what Meta's AI strategy...
RE
30 Apr