Large language models, compared

Anthropic's assistant, known for long-context reasoning, careful writing, and safety-aware outputs. Strong on analysis, coding, and document-heavy work.

Claude Fable 5 is a premium-tier Claude model priced above Opus, aimed at the hardest professional and technical work. It has been followed by Claude Fable 5.1, which is the version most teams should evaluate today.

Claude Fable 5.1 is Anthropic's premium generally available model, priced above Opus and built for the hardest reasoning, coding and analysis tasks. Anthropic says the cheaper Opus 5.5 now matches it on most work, so Fable 5.1 makes most sense where the last f

Claude Haiku 4.5 is Anthropic's fast, low-cost model at $1 and $5 per million tokens. It suits high-volume tasks such as classification, extraction and short drafting, where response speed and cost matter more than deep reasoning.

Claude Opus 4.7 is an earlier Opus release, priced at $5 and $25 per million tokens. It introduced the newer tokenizer used by later Claude models and remains available, though newer Opus models are the better starting point.

Claude Opus 4.8 was Anthropic's flagship before Opus 5 and 5.5. It is still available at $5 and $25 per million tokens and supports fast mode, but new projects will usually get better value from the newer Opus models.

Claude Opus 5, released in July 2026, was Anthropic's flagship for demanding reasoning, coding and agentic work until Claude Opus 5.5 arrived in September at a lower price. It remains a capable, widely deployed model with a 1M-token context window.

Claude Opus 5.5, released on 22 September 2026, is Anthropic's newest Opus model. Anthropic reports it performs at the level of Claude Fable 5.1 on most work while costing about 40% less to run than Opus 5, with a focus on agentic coding, computer use and know

Claude Sonnet 4.6 is the previous mid-tier Claude model, priced at $3 and $15 per million tokens. Sonnet 5 now costs less and is the better choice for new work, but 4.6 remains available for teams that need its established behavior.

Claude Sonnet 5 is the mid-tier model in the Claude 5 family and the default choice for most professional work. Anthropic made its launch pricing of $2 and $10 per million tokens permanent, which makes it one of the best price-to-quality options in the lineup.

OpenAI's flagship conversational AI. Used by professionals across every category for drafting, summarization, structured thinking, and as a Swiss-army-knife reasoning assistant.

GPT-6 Astra, released by OpenAI on 3 September 2026, is the company's most capable model to date, with state-of-the-art results on computer use, coding, cybersecurity and science. It sits at the top of the GPT-6 family in both capability and price.

GPT-6 Luna, launched on 22 September 2026, is the small, low-cost member of the GPT-6 family at $0.10 and $0.50 per million tokens. It targets high-volume tasks where cost and speed matter most.

GPT-6 Sol, launched on 22 September 2026, brings GPT-6 Astra's technology to a faster, cheaper model at $2 and $10 per million tokens. OpenAI states these are permanent prices, not promotional rates.

Google's multimodal assistant, deeply integrated into Workspace (Docs, Gmail, Sheets, Meet). Strong choice for organizations already on Google's stack.

Gemini 3.1 Pro, released by Google in February 2026, is a multimodal model with strong agentic capabilities. It is listed at $2 and $12 per million tokens for prompts up to 200,000 tokens ($4 and $18 above that) and is offered in preview on the Gemini API.

Gemini 3.8 Flash, released on 2 September 2026, is Google's newest Flash model, aimed at long-horizon engineering and agent tasks. It is priced at $0.75 and $3.75 per million tokens, a rate Google lists through 31 December 2026, after which it rises to $1.50 a

Meta's open-weights flagship model. Available for download and self-hosted deployment under the Llama license.

Mistral Large 3, released in December 2025, is Mistral AI's largest public model, a mixture-of-experts system with 675 billion total parameters of which about 41 billion are active. It is released under the Apache 2.0 license, so organizations can download and

Le Chat is Mistral AI's consumer and team assistant, built on Mistral's own models. It offers web search, document analysis and a canvas for drafting, with a free tier and European data-handling positioning.

Grok 4.7 is xAI's latest flagship model for coding, agentic tasks and knowledge work. It has a 500,000-token context window, accepts text and image input, and supports adjustable reasoning levels.

Search-first AI: every answer comes with citations. Pro tier gives access to GPT-4 / Claude / Sonar models and Pro Search (multi-step research).

Poe, from Quora, is a single chat app that gives you access to many AI models and community-built bots in one place. It suits people who want to compare answers from different models without keeping separate subscriptions.

Monica is an AI assistant that lives in your browser sidebar and works alongside whatever page you are reading. It bundles several underlying models with page summarizing, writing help and translation.

You.com is an AI search and assistant platform that combines a web-connected search engine with chat and selectable AI modes. It is aimed at people who research online and want sourced answers.

Legacy model, superseded by Gemini 3.8 Flash.

Legacy model, superseded by Gemini 3.1 Pro.

Legacy model, superseded by Mistral Large 3.

Legacy model, superseded by the GPT-6 family.

Legacy model, superseded by GPT-6 Astra, Sol and Luna.

Legacy model, superseded by the GPT-6 family.

Legacy model, superseded by Grok 4.7.