The models that write best in English

The AI Writing Benchmark measures how well leading AI models write in English. Responses to a shared corpus of English writing prompts are compared blind, without model names, and the verdicts are combined into a ranking for English.

Leading across all categories

The 10 best models

The ten best models for writing in English
#	Model	Score
1	Claude Opus 5 💡 adaptive high	93.9
2	Kimi K3 💡 enabled	93.5
3	Claude Fable 5 💡 adaptive high	93.1
4	Claude Fable 5.1 💡 adaptive high	92.4
4	GPT-6 Astra 💡 high	92.4
6	GLM 5.3 💡 max	90.7
7	Claude Opus 4.8 💡 adaptive high	89.4
8	Claude Sonnet 5 💡 adaptive high	86.4
9	GPT-5.6 Sol 💡 high	85.9
10	GLM 5.3 Flash 💡 high	83.8

This score is not a percentage grade: 50 means an estimated one-in-two chance of beating the average model in English.

Last updated: September 14, 2026 at 3:13 a.m. EDT

See the full leaderboard

The 10 best local variants

Estimated VRAM is the approximate graphics memory needed to load the entire model.

The ten best exact variants run locally
#	Model	Score	Estimated VRAM
1	Muse Glimmer 30B K-Quant Dynamic 💡 high 🏠	76.7	≈ 32 GB
2	Qwen3.8 27B Q4_K_M 💡 xhigh** 🏠	66.5	≈ 24 GB
3	Muse Glimmer 30B K-Quant 17GB 💡 high 🏠	64.2	≈ 24 GB
4	Qwen3.6 27B Q8_0 💡 on* 🏠	62.4	≈ 32 GB
5	Qwen3.8 27B Q8_0 💡 xhigh** 🏠	61.1	≈ 48 GB
6	Gemma 4 26B A4B Q6_K 💡 on* 🏠	57.6	≈ 32 GB
7	Gemma 4 31B Q6_K 💡 on* 🏠	57.4	≈ 32 GB
8	Qwen3.8 27B Q6_K 💡 xhigh** 🏠	56.4	≈ 32 GB
9	Gemma 4 31B Q8_0 💡 on* 🏠	56.1	≈ 48 GB
10	Qwen3.6 27B Q4_K_M 💡 on* 🏠	55.7	≈ 24 GB

Some software can keep part of the model in system memory. This reduces the VRAM required, but usually makes generation slower.

90models compared

54586published AI verdicts

The ranking evaluates writing quality alone. Prompts are written directly in the language being tested, and model names remain hidden until the verdict so reputation cannot influence the evaluation. This site is a personal project. No model provider pays me, and I have no reason to favour one model over another.

Explore the results

Published 12 August 2026

Muse Glimmer 30B writing test: how it compares to local rivals

Our updated writing test compares both Muse Glimmer quantizations with Gemma, Qwen and Nemotron across four languages, then weighs memory, speed and factual risk.

Published 29 July 2026

Does AI Write Better in English With or Without Reasoning?

In the current direct sample, reasoning-enabled responses earned 66.2% of the available points in English. The uncertainty interval still includes parity, the result differs sharply from one model to the next, and no human votes have compared the two settings directly yet.