AI models
Model families: lineup, access, fit and limits, with a dated spec table.
- Model · Anthropic · re-verifyClaude (Anthropic)
Claude is Anthropic's model family. Its current Fable 5.1 and Opus, Sonnet and Haiku 5.5 models accept text and images and return text, with tool support and a 1M-token context window. Choose by measured task quality, latency and total cost rather than the family name alone.
Verified Oct 18, 2026 · SOPs, Research, Data - Model · Cohere · re-verifyCohere models (Command, Embed, Rerank)
Cohere builds models for enterprise retrieval and generation: the Command family for chat and agents, Embed and Rerank for search, plus parsing, transcription and multilingual models. It is available on its own platform and through major clouds, which matters if procurement wants AI inside an existing cloud contract.
Verified Oct 18, 2026 · Research, Support, Build - Model · DeepSeek · re-verifyDeepSeek models
DeepSeek offers two API models: deepseek-flash for low-cost volume work and deepseek-v4-pro for harder tasks. Both have a 1M-token context window, thinking and non-thinking modes, JSON output and tool calls, and off-peak use is half price. DeepSeek's privacy policy says personal data is stored in China, so review it before you send anything sensitive.
Verified Oct 18, 2026 · Build, Data - Model · Google · re-verifyGemini models (Google)
Gemini is Google's model family, available through the Gemini app, Google Workspace and the Gemini API. Google's current flagship reasoning model is Gemini 4 Argon, with a 1M-token context window. The Flash line covers fast everyday and agentic work. Sister models handle images (Nano Banana), video (Veo), music (Lyria) and embeddings.
Verified Oct 18, 2026 · Research, Data, Content - Model · Google · re-verifyGemma 4
Google’s Gemma 4 family provides model weights for deployment in a compatible runtime. Choose the exact variant, check its supported inputs and test it on your own examples; the model card is not a guarantee of performance in a particular application.
Verified Oct 18, 2026 · Local, Evals - Model · OpenAI · re-verifyGPT models (OpenAI)
OpenAI's GPT models power ChatGPT and the OpenAI API. OpenAI currently points to GPT-6 Astra for complex reasoning and coding, GPT-6.1 Sol to balance capability and cost, and GPT-6 Luna for high-volume work. All three list a 1.05M-token context window and built-in web search, file search and computer-use tools.
Verified Oct 18, 2026 · Content, SOPs, Research - Model · Mistral AI · re-verifyMistral models (Mistral AI)
Mistral AI publishes both open-weight and commercial models. The current lineup spans frontier multimodal models (Mistral Large 4 and Medium 3.5), small Apache 2.0 models (Mistral Small 4, Ministral 3), and specialists for code, OCR, speech, embeddings and moderation. Pick it when you need weights you can host yourself or a European vendor.
Verified Oct 18, 2026 · Build, Research - Model · Perplexity · re-verifyPerplexity Sonar and Agent API
Perplexity's Agent API gives developers one interface to models from several providers, plus Perplexity's own search-grounded Sonar model. Perplexity ended support for the older Sonar Chat Completions API on September 27, 2026, and points users to Agent API presets instead. Sonar is listed at $0.25 input and $2.50 output per million tokens.
Verified Oct 18, 2026 · Research, Content - Model · Microsoft · re-verifyPhi models (Microsoft)
Phi is Microsoft's open model family. The MIT-licensed Phi-4-mini-instruct is a 3.8B text model; Phi-4-reasoning-vision-15B accepts text and images. Compare the exact model, runtime, hardware and task before choosing local or hosted deployment.
Verified Oct 18, 2026 · Build - Model · Alibaba (Qwen team) · re-verifyQwen models (Alibaba)
Qwen is a model family from Alibaba's Qwen team. Some models are released with open weights on Hugging Face, such as Qwen3.8-27B under Apache 2.0. Hosted models, including Qwen3.8-Max and Qwen3.8-Flash, are served through Qwen Cloud, which the Qwen team calls its official API. Review data terms before business use.
Verified Oct 18, 2026 · Build, Content
Other directories: Needs & problems · Questions & how-tos · Services & tools · Concepts & glossary · Use-case stacks