Skip to content
fedi.software

starling-lm-7b-alpha in fedi-index

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

fedi-index · General

D 1,116 ±5

Percentile 14 · #252 · provisional

People prefer / Solves tasks

1,110 / —

Human votes vs benchmarks, Elo-eq

Value

—

On your hardware

Closed weights: cloud only.

Score decomposition

D General 1,116 [1,112–1,121] Percentile 14 · #252 · components: 6 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/overall 1,108 ±4.04 1,108 ±4 28.0% 2026-10-02 —
LMArena text/hard_prompts 1,075 ±6.38 1,102 ±6 26.2% 2026-10-02 —
LMArena text/instruction_following 1,070 ±5.76 1,100 ±6 13.2% 2026-10-02 —
LMArena text/expert 1,033 ±16 1,094 ±13 8.7% 2026-10-02 —
LMArena text/multi_turn 1,085 ±8.86 1,113 ±8 5.8% 2026-10-02 —
LMArena text/longer_query 1,081 ±12 1,103 ±11 5.0% 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 1,122 [1,109–1,135] Percentile 12 · #244 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/coding 1,098 ±8.27 1,112 ±7 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 Math 1,103 [1,090–1,116] Percentile 9 · #231 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,081 ±7.96 1,088 ±8 86.4% 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 1,122 [1,111–1,134] Percentile 14 · #232 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,099 ±7.81 1,121 ±8 64.8% 2026-10-02 —
LMArena text/longer_query 1,081 ±12 1,103 ±11 27.4% 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 1,101 [1,085–1,116] Percentile 6 · #209 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,045 ±11 1,092 ±9 92.3% 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 1,118 [1,109–1,127] Percentile 15 · #232 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,139 ±4.57 1,112 ±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: Русский 1,140 [1,118–1,162] Percentile 9 · #214 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,096 ±14 1,133 ±12 90.5% 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

History

fedi-index by month
The same in a table
Model 2026-10
starling-lm-7b-alpha 1,116 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: starling-lm-7b-alpha

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