Skip to content
fedi.software

phi-3-mini-128k-instruct in fedi-index

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

fedi-index · General

D 1,070 ±4

Percentile 6 · #275 · provisional

People prefer / Solves tasks

1,061 / —

Human votes vs benchmarks, Elo-eq

Value

-272

$0.2275 per 1M tokens, median of 1 providers

On your hardware

Closed weights: cloud only.

Score decomposition

D General 1,070 [1,066–1,074] Percentile 6 · #275 · components: 6 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/overall 1,050 ±3.68 1,050 ±4 26.8% 2026-10-02 —
LMArena text/hard_prompts 1,028 ±5.47 1,059 ±5 25.7% 2026-10-02 —
LMArena text/instruction_following 1,023 ±5.16 1,056 ±5 12.8% 2026-10-02 —
LMArena text/expert 984 ±10 1,054 ±8 10.8% 2026-10-02 —
LMArena text/multi_turn 989 ±7.92 1,026 ±7 5.8% 2026-10-02 —
LMArena text/longer_query 996 ±8.05 1,023 ±8 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 1,075 [1,065–1,085] Percentile 6 · #260 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/coding 1,039 ±6.53 1,060 ±6 89.3% 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 Math 1,109 [1,098–1,120] Percentile 11 · #228 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,089 ±6.70 1,095 ±6 87.1% 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 Writing 1,049 [1,039–1,059] Percentile 5 · #255 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,024 ±7.11 1,044 ±7 62.2% 2026-10-02 —
LMArena text/longer_query 996 ±8.05 1,023 ±8 30.7% 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 Language: 繁體中文 1,078 [1,066–1,090] Percentile 4 · #215 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,016 ±8.10 1,069 ±6 93.1% 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: Deutsch 1,075 [1,055–1,094] Percentile 1 · #142 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/german 1,006 ±13 1,062 ±11 91.3% 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: English 1,063 [1,054–1,072] Percentile 7 · #255 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,083 ±4.56 1,053 ±5 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 1,097 [1,069–1,125] Percentile 1 · #148 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/spanish 1,058 ±18 1,082 ±16 88.1% 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: Français 1,072 [1,044–1,101] Percentile 1 · #134 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/french 1,038 ±18 1,055 ±16 87.9% 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: 日本語 1,066 [1,045–1,088] Percentile 0 · #128 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/japanese 899 ±16 1,052 ±12 90.6% 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: Русский 1,062 [1,048–1,075] Percentile 2 · #230 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,004 ±8.06 1,051 ±7 92.8% 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-128k-instruct 1,070 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-128k-instruct

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