Fictionlivebench — Epoch AI

Fictionlivebench — Epoch AI is run by Epoch AI. It has ranked 42 models, of which the catalogue holds 30, scoring from 0.188 to 1.

measured by Epoch AI · the board itself

Place	Model	Metric	Score
1stof 42	o3	120k token score (medium)	1
2ndof 42	GPT-5	120k token score (medium)	0.969
5thof 42	o3-pro	120k token score (medium)	0.889
6thof 42	Gemini 2.5 Pro Preview 06-05	120k token score	0.875
7thof 42	Kimi K2.5	120k token score	0.781
8thof 42	Grok 4 Fast	120k token score	0.75
9thof 42	Gemini 2.5 Pro Preview 05-06	120k token score	0.719
10thof 42	Gemini 2.5 Pro	120k token score	0.719
11thof 42	Qwen3 235B A22B Thinking 2507	120k token score	0.688
12thof 42	Gemini 2.5 Flash	120k token score	0.688
14thof 42	Grok 3 Mini	120k token score (medium)	0.656
16thof 42	Qwen3 Next 80B A3B Instruct	120k token score	0.625
17thof 42	gemini-2.0-flash-001	120k token score	0.625
18thof 42	GPT-4.1 fine-tuned	120k token score	0.625
19thof 42	GPT-5 mini	120k token score (medium)	0.625
20thof 42	o4-mini	120k token score (medium)	0.625
21stof 42	MiniMax M1	120k token score	0.594
22ndof 42	Grok 3	120k token score	0.583
23rdof 42	Claude 3.7 Sonnet	120k token score	0.531
24thof 42	o1	120k token score (medium)	0.531
25thof 42	DeepSeek-V3.1	120k token score	0.531
28thof 42	GPT-4.1 mini	120k token score	0.469
29thof 42	parasail-qwen3-235b-a22b-instruct-2507	120k token score	0.444
30thof 42	o3-mini	120k token score (medium)	0.438
32ndof 42	Claude Opus 4	120k token score	0.375
35thof 42	Llama-4-Maverick-17B-128E-Instruct	120k token score	0.364
36thof 42	Claude Sonnet 4	120k token score	0.364
38thof 42	DeepSeek-R1	120k token score	0.333
41stof 42	GPT-5 nano	120k token score (medium)	0.219
42ndof 42	GPT-4.1 nano	120k token score	0.188

Place

Model

Metric

Score

1stof 42

120k token score (medium)

2ndof 42

120k token score (medium)

0.969

5thof 42