Skip to content
fedi.software

phi-3-mini-4k-instruct in fedi-index

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

fedi-index · General

D 1,099 ±4

Percentile 10 · #265 · provisional

People prefer / Solves tasks

1,092 / —

Human votes vs benchmarks, Elo-eq

Value

-243

$0.2275 per 1M tokens, median of 1 providers

On your hardware

Closed weights: cloud only.

Score decomposition

D General 1,099 [1,095–1,102] Percentile 10 · #265 · components: 6 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/overall 1,073 ±3.21 1,073 ±3 26.7% 2026-10-02 —
LMArena text/hard_prompts 1,072 ±4.82 1,099 ±4 25.7% 2026-10-02 —
LMArena text/instruction_following 1,053 ±4.45 1,085 ±4 12.9% 2026-10-02 —
LMArena text/expert 1,045 ±9.86 1,104 ±8 10.8% 2026-10-02 —
LMArena text/multi_turn 1,018 ±6.81 1,052 ±6 6.0% 2026-10-02 —
LMArena text/longer_query 1,044 ±7.62 1,068 ±7 5.7% 2026-10-02 —

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

D Coding 1,117 [1,108–1,126] Percentile 12 · #246 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/coding 1,093 ±5.92 1,107 ±5 89.5% 2026-10-02 —

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

D Math 1,127 [1,117–1,137] Percentile 13 · #222 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,111 ±6.23 1,116 ±6 87.4% 2026-10-02 —

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

D Writing 1,071 [1,062–1,080] Percentile 7 · #250 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,037 ±6.44 1,058 ±7 62.6% 2026-10-02 —
LMArena text/longer_query 1,044 ±7.62 1,068 ±7 30.4% 2026-10-02 —

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

D Language: 繁體中文 1,082 [1,071–1,092] Percentile 4 · #214 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,021 ±7.36 1,073 ±6 93.3% 2026-10-02 —

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

D Language: Deutsch 1,104 [1,085–1,123] Percentile 4 · #139 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/german 1,044 ±12 1,095 ±10 91.5% 2026-10-02 —

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

D Language: English 1,096 [1,088–1,104] Percentile 11 · #243 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,117 ±4.13 1,089 ±4 93.6% 2026-10-02 —

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

D Language: Español 1,118 [1,091–1,144] Percentile 2 · #146 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/spanish 1,085 ±17 1,107 ±15 88.6% 2026-10-02 —

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

D Language: Français 1,103 [1,074–1,132] Percentile 2 · #132 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/french 1,075 ±19 1,089 ±17 87.5% 2026-10-02 —

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

D Language: 日本語 1,091 [1,068–1,114] Percentile 1 · #127 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/japanese 935 ±17 1,079 ±13 90.1% 2026-10-02 —

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

D Language: 한국어 1,071 [1,054–1,088] Percentile 2 · #134 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/korean 905 ±13 1,059 ±10 91.9% 2026-10-02 —

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

D Language: Русский 1,076 [1,065–1,087] Percentile 3 · #227 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,022 ±6.67 1,067 ±6 93.2% 2026-10-02 —

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
phi-3-mini-4k-instruct 1,099 D

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: phi-3-mini-4k-instruct

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