Skip to content
fedi.software

DeepSeek V4.1 Flash in fedi-index

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

fedi-index · General

A 1,448 ±10

Percentile 90 · #31

People prefer / Solves tasks

1,452 / 1,441

Human votes vs benchmarks, Elo-eq

Value

+92◆

$0.42 per 1M tokens, median of 47 providers · vendor $0.2625

On your hardware

Does not fit the reference setups.

Score decomposition

A General 1,448 [1,438–1,459] Percentile 90 · #31 · components: 7
Source Component Value Elo-eq Weight Measured Variant
Epoch AI eci 155 ±2.26 1,476 ±18 28.8% 2026-09-09 —
LMArena text/overall 1,462 ±3.52 1,462 ±4 18.8% 2026-10-02 max
LMArena text/hard_prompts 1,483 ±4.17 1,470 ±4 18.6% 2026-10-02 max
LMArena text/instruction_following 1,476 ±5.51 1,486 ±5 8.8% 2026-10-02 max
LMArena text/expert 1,504 ±9.41 1,479 ±8 7.8% 2026-10-02 max
LMArena text/longer_query 1,474 ±5.03 1,470 ±5 4.5% 2026-10-02 max
LMArena text/multi_turn 1,453 ±8.54 1,447 ±8 3.9% 2026-10-02 max

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

A Coding 1,452 [1,448–1,457] Percentile 94 · #18 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena webdev/overall 1,620 ±5.33 1,484 ±2 47.6% 2026-10-01 max
LMArena text/coding 1,507 ±6.10 1,476 ±5 42.4% 2026-10-02 max

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 Agents 1,449 [1,447–1,451] Percentile 72 · #17 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena agent/overall 0.04 ±0.00 1,475 ±1 90.5% 2026-10-02 max

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

A Math 1,426 [1,405–1,448] Percentile 84 · #42 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,486 ±14 1,476 ±13 81.9% 2026-10-02 max

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

A Writing 1,448 [1,439–1,458] Percentile 85 · #41 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,436 ±7.47 1,465 ±8 59.1% 2026-10-02 max
LMArena text/longer_query 1,474 ±5.03 1,470 ±5 33.9% 2026-10-02 max

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

A Language: 繁體中文 1,435 [1,419–1,452] Percentile 81 · #43 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,502 ±11 1,455 ±9 92.1% 2026-10-02 max

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

A Language: English 1,451 [1,440–1,461] Percentile 90 · #28 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,476 ±5.31 1,468 ±6 93.4% 2026-10-02 max

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

A Language: Русский 1,450 [1,436–1,465] Percentile 88 · #29 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,470 ±8.76 1,470 ±8 92.6% 2026-10-02 max

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
DeepSeek V4.1 Flash 1,448 A

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: DeepSeek V4.1 Flash

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