Skip to content
fedi.software

Claude Sonnet 4 in fedi-index

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

fedi-index · General

B 1,359 ±6

Percentile 51 · #145

People prefer / Solves tasks

1,359 / 1,359

Human votes vs benchmarks, Elo-eq

Value

-62

$6 per 1M tokens, median of 14 providers

On your hardware

Closed weights: cloud only.

Score decomposition

B General 1,359 [1,353–1,366] Percentile 51 · #145 · components: 8
Source Component Value Elo-eq Weight Measured Variant
Epoch AI eci 142 ±1.04 1,372 ±8 37.5% 2025-05-22 —
LMArena text/overall 1,351 ±2.30 1,351 ±2 15.9% 2026-10-02 thinking-32k
LMArena text/hard_prompts 1,372 ±2.93 1,370 ±3 15.8% 2026-10-02 thinking-32k
LMArena text/instruction_following 1,376 ±3.60 1,390 ±3 7.7% 2026-10-02 thinking-32k
LMArena text/expert 1,371 ±7.47 1,371 ±6 7.0% 2026-10-02 thinking-32k
LMArena text/longer_query 1,398 ±3.93 1,398 ±4 3.8% 2026-10-02 thinking-32k
LMArena text/multi_turn 1,376 ±4.27 1,378 ±4 3.8% 2026-10-02 thinking-32k
Epoch AI gpqa_diamond decay ×0.26 0.79 ±0.03 1,408 ±12 1.3% 2025-05-26 59k

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 1,375 [1,368–1,381] Percentile 54 · #127 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/coding 1,414 ±4.03 1,394 ±4 90.1% 2026-10-02 thinking-32k

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 Math 1,348 [1,337–1,358] Percentile 46 · #138 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,375 ±6.60 1,369 ±6 87.2% 2026-10-02 thinking-32k

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 Vision 1,362 [1,350–1,374] Percentile 33 · #76 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena vision/overall 1,191 ±8.64 1,376 ±8 76.2% 2026-10-02 thinking-32k
LMArena vision/ocr 1,196 ±11 1,371 ±11 16.2% 2026-10-02 thinking-32k

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 1,370 [1,364–1,376] Percentile 54 · #123 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,345 ±4.86 1,372 ±5 61.4% 2026-10-02 thinking-32k
LMArena text/longer_query 1,398 ±3.93 1,398 ±4 32.2% 2026-10-02 thinking-32k

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: 繁體中文 1,326 [1,317–1,336] Percentile 32 · #152 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,351 ±6.34 1,335 ±5 93.5% 2026-10-02 thinking-32k

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: Deutsch 1,332 [1,317–1,347] Percentile 29 · #103 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/german 1,330 ±9.58 1,342 ±8 92.5% 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

C Language: English 1,341 [1,335–1,347] Percentile 44 · #154 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,364 ±2.97 1,350 ±3 93.9% 2026-10-02 thinking-32k

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 1,346 [1,327–1,364] Percentile 20 · #119 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/spanish 1,357 ±11 1,359 ±10 91.7% 2026-10-02 thinking-32k

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 1,340 [1,318–1,361] Percentile 17 · #112 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/french 1,364 ±13 1,353 ±12 90.8% 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

C Language: 日本語 1,342 [1,328–1,356] Percentile 28 · #93 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/japanese 1,301 ±10 1,353 ±8 92.7% 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: 한국어 1,342 [1,328–1,357] Percentile 24 · #104 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/korean 1,292 ±11 1,353 ±8 92.6% 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: Polski 1,334 [1,325–1,343] Percentile 13 · #114 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/polish 1,324 ±5.80 1,343 ±5 93.5% 2026-10-02 thinking-32k

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: Русский 1,356 [1,346–1,365] Percentile 41 · #138 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,355 ±6.06 1,366 ±5 93.4% 2026-10-02 thinking-32k

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
Claude Sonnet 4 1,359 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: Claude Sonnet 4

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