Lech mazur writing — Epoch AI

Lech mazur writing — Epoch AI is run by Epoch AI. It has ranked 49 models, of which the catalogue holds 38, scoring from 6.05 to 8.6.

measured by Epoch AI · the board itself

Place	Model	Metric	Score
1stof 49	GPT-5	Mean score (medium)	8.6
2ndof 49	Kimi K2 Instruct	Mean score	8.56
3rdof 49	Claude Opus 4.1	Mean score	8.47
5thof 49	o3-pro	Mean score (medium)	8.44
6thof 49	o3	Mean score (medium)	8.39
7thof 49	Gemini 2.5 Pro	Mean score	8.38
8thof 49	Claude Opus 4	Mean score	8.36
10thof 49	GPT-5 mini	Mean score (medium)	8.31
11thof 49	Qwen3 235B A22B	Mean score	8.3
12thof 49	DeepSeek-R1	Mean score	8.3
13thof 49	Qwen3 235B A22B Thinking 2507	Mean score	8.24
14thof 49	DeepSeek-R1 (0528)	Mean score	8.19
15thof 49	GPT-4o (2024-11-20)	Mean score	8.18
16thof 49	Claude Sonnet 4	Mean score	8.14
17thof 49	Claude 3.7 Sonnet	Mean score	8.11
19thof 49	Gemini 2.5 Pro Preview 05-06	Mean score	8.09
21stof 49	Claude 3.5 Sonnet	Mean score	8.03
22ndof 49	QwQ-32B	Mean score	8.02
23rdof 49	Gemma 3 27B	Mean score	7.99
25thof 49	Mistral Medium 2505	Mean score	7.73
26thof 49	GPT OSS 120B	Mean score	7.71
27thof 49	DeepSeek-V3 0324	Mean score	7.7
29thof 49	Gemini 2.5 Flash	Mean score	7.65
30thof 49	Grok 3	Mean score	7.64
32ndof 49	Qwen3 30B A3B	Mean score	7.53
33rdof 49	o4-mini	Mean score (medium)	7.5
35thof 49	Claude Haiku 3.5	Mean score	7.35
36thof 49	Grok 3 Mini	Mean score (low)	7.35
37thof 49	GLM-4.5	Mean score	7.34
39thof 49	Gemini 2.0 Flash	Mean score	7.15
40thof 49	o1	Mean score (medium)	7.02
41stof 49	Mistral Large 2407	Mean score	6.9
42ndof 49	GPT-4o-mini (2024-07-18)	Mean score	6.72
43rdof 49	o1-mini	Mean score (medium)	6.49
45thof 49	Phi-4	Mean score	6.26
46thof 49	Llama-4-Maverick-17B-128E-Instruct	Mean score	6.2
47thof 49	o3-mini	Mean score (high)	6.17
49thof 49	Nova Pro	Mean score	6.05

Place

Model

Metric

Score

1stof 49

Mean score (medium)

8.6

2ndof 49

Mean score

8.56