Notus is a 7B parameter model from Argilla that consistently outperforms Zephyr on chat benchmarks using preference optimization…
Starling LM from Berkeley is an open-source chat model trained with RLAIF (Reinforcement Learning from AI Feedback) that…
Orca 2 by Microsoft Research teaches small models advanced reasoning through imitation learning from larger models. 7B and…
Solar Mini is a cost-efficient small model from Upstage optimized for document understanding and RAG applications. Processes long…
Mixtral 8x7B by Mistral AI uses a Mixture of Experts architecture to achieve GPT-3.5 performance with lower latency.…
Stable LM by Stability AI is a family of open-source language models ranging from 3B to 12B parameters.…
WizardLM uses Evol-Instruct methodology to create highly capable instruction-following models. Dramatically improves base model capability by training on…
Gemma is Google's family of open-source lightweight models designed for running locally on laptops and edge devices. Strong…
Microsoft's Phi-3 is a small but powerful open-source language model that runs efficiently on mobile devices and CPUs.…
Qwen by Alibaba is a powerful open-source language model family with versions from 1.5B to 72B parameters. Strong…
DeepSeek provides frontier open-source AI models at a fraction of the cost of competitors. DeepSeek-V3 matches GPT-4o performance…
Yi by 01.ai (founded by Kai-Fu Lee) is a bilingual Chinese-English language model competitive with leading frontier models.…