Topic
Models
Language models, open weights, and the systems built from them.
7 entries in Super Intelligence
Models
GPT-5
GPT-5 is OpenAI's unified model family, released in August 2025, which routes requests between a fast general model and a slower reasoning model. It covers the system's design, versions, benchmarks, pricing, and disputes.
Models
Claude
Claude is Anthropic's family of large language models, released from March 2023 and sold in Haiku, Sonnet and Opus tiers. It is trained with constitutional AI and promoted as a safety-focused alternative to rival chatbots.
Models
Large language model
A large language model is a neural network trained on vast text corpora to predict the next token, giving it broad but unreliable competence in language, code, and reasoning tasks.
Models
Gemini
Gemini is Google DeepMind's family of multimodal models, announced in December 2023 and built into Search, Workspace, Android and the Gemini app. It spans small on-device models and large reasoning systems.
Models
Llama
Llama is Meta's family of open-weight language models, first released in February 2023. Its downloadable weights made capable models widely available and turned licensing into a central argument in AI policy.
Models
Qwen
Qwen is Alibaba's family of language and multimodal models, released from 2023 under the Tongyi Qianwen name. It is among the most widely used open-weight families, with versions from phones to trillion-parameter systems.
Models
Open-source AI
Open-source AI covers systems whose code, weights, or training data are published under licences that permit reuse, modification, and redistribution, and the debate over how open they really are.