The Best Large Language Model for x

Last updated April 23, 2026.

General chat, breadth of knowledge in the humanities and STEM: Claude Opus 4.7, due to its excellent reasoning traces and post-trained personality. Opus 4.7 is the best model for knowledge work. Runner-up: GPT-5.5, for its accuracy and speed. No current model is as fast as GPT-5.5. Its personality is quite dry, however.

Coding, agentic work, and tool-calling through the native harnesses: Claude Opus 4.7, due to its exceptional post-training and coding style. Runner up: GPT-5.5, due to the great Codex harness and depth of knowledge.

Web search: Claude Opus 4.7, due to its proficiency in reasoning over the data it finds on the web. Runner-up: GPT-5.5 in ChatGPT, by virtue of Gemini’s incompetence.

Fact-based medium- to long-context retrieval (PDFs and plain text): Gemini 3.1 Pro, due to its massive context window and native multimodality. Runner-up: Claude Opus 4.7, due to its amazing reasoning traces.*

Reasoning-based medium- to long-context analysis (PDFs and plain text): Claude Opus 4.7, due to its reasoning. Runner-up: Gemini 3.1 Pro, due to its context window and multimodality.*

Image analysis: Claude Opus 4.7, due to its reasoning — what it does with those images. Runner-up: Gemini 3.1 Pro and GPT-5.5 are tied and very close to Opus 4.7. Both models are natively multimodal, but Gemini provides more inference / image uploads every five hours on the $20 plan.

* Gemini is better at retrieving and interpreting data across long contexts, while Claude is better at reasoning over it. Use Claude when you need to work with it to accomplish a difficult task; use Gemini for “fancy Command-F.” Both are equivalent in PDF analysis, so long as the PDFs are pre-OCR’d — otherwise, Gemini has the edge. Both models are more proficient than GPT-5.4 in ChatGPT.