Skip to content
fedi.software

GPT-5.4 Mini in fedi-index

The score in every slice, what it is made of, value, local run and history.

fedi-index · General

B 1,405 ±5

Percentile 71 · #86

People prefer / Solves tasks

1,403 / 1,408

Human votes vs benchmarks, Elo-eq

Value

+14

$1.69 per 1M tokens, median of 31 providers · vendor $1.69

On your hardware

Closed weights: cloud only.

Score decomposition

B General 1,405 [1,400–1,410] Percentile 71 · #86 · components: 9
Source Component Value Elo-eq Weight Measured Variant
Epoch AI eci 149 ±0.82 1,429 ±6 34.7% 2026-03-17 —
LMArena text/overall 1,411 ±1.90 1,411 ±2 15.4% 2026-10-02 high
LMArena text/hard_prompts 1,424 ±2.27 1,417 ±2 15.4% 2026-10-02 high
LMArena text/instruction_following 1,405 ±2.76 1,418 ±3 7.6% 2026-10-02 high
LMArena text/expert 1,435 ±4.18 1,423 ±3 7.4% 2026-10-02 high
Epoch AI simpleqa_verified decay ×0.61 0.29 ±0.01 1,390 ±4 3.8% 2026-08-27 high
LMArena text/longer_query 1,407 ±2.67 1,407 ±2 3.8% 2026-10-02 high
LMArena text/multi_turn 1,429 ±3.39 1,425 ±3 3.8% 2026-10-02 high
Epoch AI gpqa_diamond decay ×0.26 0.87 ±0.02 1,444 ±11 1.2% 2026-08-07 xhigh

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

B Coding 1,397 [1,395–1,400] Percentile 67 · #93 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena webdev/overall 1,397 ±3.32 1,420 ±1 45.8% 2026-10-01 high
LMArena text/coding 1,438 ±2.99 1,415 ±3 44.6% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

B Math 1,401 [1,395–1,408] Percentile 72 · #71 · components: 3
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,418 ±5.47 1,411 ±5 51.4% 2026-10-02 high
Epoch AI frontiermath_t1_3 decay ×0.86 0.51 ±0.03 1,432 ±6 28.1% 2026-06-12 xhigh
Epoch AI frontiermath_t4 decay ×0.86 0.10 ±0.05 1,408 ±8 13.3% 2026-06-12 xhigh

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

B Vision 1,413 [1,409–1,417] Percentile 54 · #53 · components: 3 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena vision/overall 1,245 ±3.17 1,425 ±3 63.0% 2026-10-02 high
LMArena vision/ocr 1,253 ±3.18 1,426 ±3 15.7% 2026-10-02 high
LMArena vision/diagram 1,266 ±4.43 1,433 ±4 15.3% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

B Writing 1,389 [1,384–1,393] Percentile 61 · #106 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,370 ±3.55 1,398 ±4 61.9% 2026-10-02 high
LMArena text/longer_query 1,407 ±2.67 1,407 ±2 31.9% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

B Language: 繁體中文 1,397 [1,390–1,405] Percentile 54 · #102 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,446 ±5.18 1,410 ±4 93.7% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

B Language: Deutsch 1,392 [1,376–1,409] Percentile 52 · #69 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/german 1,407 ±10 1,408 ±9 92.2% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

B Language: English 1,389 [1,384–1,394] Percentile 58 · #114 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,412 ±2.51 1,401 ±3 94.0% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

C Language: Español 1,389 [1,376–1,402] Percentile 40 · #89 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/spanish 1,405 ±7.56 1,403 ±7 92.9% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

B Language: Français 1,408 [1,395–1,420] Percentile 50 · #68 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/french 1,440 ±7.39 1,423 ±7 93.0% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

C Language: 日本語 1,391 [1,374–1,409] Percentile 46 · #69 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/japanese 1,375 ±13 1,408 ±10 91.8% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

C Language: 한국어 1,396 [1,382–1,410] Percentile 49 · #70 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/korean 1,368 ±10 1,411 ±8 92.7% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

C Language: Polski 1,405 [1,391–1,418] Percentile 48 · #69 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/polish 1,415 ±8.86 1,420 ±8 92.8% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

B Language: Русский 1,408 [1,402–1,415] Percentile 66 · #80 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,417 ±4.04 1,422 ±4 93.8% 2026-10-02 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

History

fedi-index by month
The same in a table
Model 2026-10
GPT-5.4 Mini 1,405 B

Badge

Show the model's fedi-index on your site or in a README. The badge updates itself; it links to this page.

fedi-index: GPT-5.4 Mini

Data CC BY 4.0: keep the link to fedi.software.