# Best AI model for your job

> Every job has a permanent page with the current shortlist, the evidence it rests on, and the checks to run before choosing.

## Write or improve something

- [Best AI model for emails](/best/for/emails) — MAI-Thinking-1 · 95.4
- [Best AI model for copywriting](/best/for/marketing) — MAI-Thinking-1 · 95.4
- [Best AI model for creative writing](/best/for/fiction) — MAI-Thinking-1 · 95.4
- [Best AI model for roleplay](/best/for/roleplay) — specialist evidence

## Learn, research, or solve a problem

- [Best AI model for studying](/best/for/studying) — Claude Fable 5.1 · 86.7
- [Best AI model for research](/best/for/research) — Claude Fable 5.1 · 86.7
- [Best AI model for reasoning](/best/for/reasoning) — GPT-6 Astra · 89.5
- [Best AI model for math](/best/for/math) — Kimi K2.6 · 71.0
- [Best AI model for legal research](/best/for/legal) — specialist evidence
- [Best AI model for medical research](/best/for/medical) — specialist evidence

## Work with documents or data

- [Best AI model for Excel](/best/for/spreadsheets) — Claude Fable 5.1 · 83.8
- [Best AI model for data analysis](/best/for/data-analysis) — Claude Fable 5.1 · 83.8
- [Best AI model for data extraction](/best/for/extraction) — MAI-Thinking-1 · 95.4
- [Best AI model for presentations](/best/for/presentations) — Claude Fable 5.1 · 80.2

## Build or fix software

- [Best AI model for coding](/best/for/coding) — Claude Fable 5.1 · 83.8
- [Best AI model for debugging](/best/for/debugging) — Claude Fable 5.1 · 83.8
- [Best AI model for website building](/best/for/frontend) — Claude Fable 5.1 · 83.8
- [Best local LLM for coding](/best/for/local-code) — GLM-5.3 · 61.4

## Build an AI-powered workflow

- [Best AI model for agents](/best/for/agents) — Claude Fable 5.1 · 80.2
- [Best AI model for RAG](/best/for/rag) — GPT-6 Astra · 89.5
- [Best AI model for customer service](/best/for/support) — Claude Fable 5.1 · 80.2

## Work across languages

- [Best AI model for translation](/best/for/translation) — Qwen3.7 Max · 100.0
- [Best AI model for language learning](/best/for/language-learning) — Qwen3.7 Max · 100.0

## Work with voice, images, or video

- [Best AI voice assistant](/best/for/voice) — specialist evidence
- [Best AI model for meeting notes](/best/for/transcription) — specialist evidence
- [Best AI model for images](/best/for/images) — specialist evidence
- [Best AI model for video](/best/for/video) — specialist evidence

## Find a useful everyday assistant

- [Best AI assistant](/best/for/general) — Claude Fable 5.1 · 84.7

Data updated September 18, 2026. Interactive selector: https://benchlm.ai/tools/llm-selector

Canonical page: https://benchlm.ai/best/for