Skip to content
fedi.software

DeepSeek V3.1 in fedi-index

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

fedi-index · General

B 1,377 ±10

Percentile 56 · #130

People prefer / Solves tasks

1,399 / 1,340

Human votes vs benchmarks, Elo-eq

Value

+19

$0.4525 per 1M tokens, median of 19 providers

On your hardware

Does not fit the reference setups.

Score decomposition

B General 1,377 [1,367–1,388] Percentile 56 · #130 · components: 7
Source Component Value Elo-eq Weight Measured Variant
Epoch AI eci 140 ±1.97 1,358 ±16 32.2% 2025-08-21 —
LMArena text/overall 1,420 ±3.09 1,420 ±3 18.4% 2026-10-02 —
LMArena text/hard_prompts 1,417 ±4.40 1,410 ±4 17.9% 2026-10-02 thinking
LMArena text/instruction_following 1,400 ±5.67 1,413 ±5 8.5% 2026-10-02 thinking
LMArena text/expert 1,403 ±13 1,397 ±10 6.5% 2026-10-02 thinking
LMArena text/longer_query 1,422 ±6.11 1,421 ±6 4.2% 2026-10-02 thinking
LMArena text/multi_turn 1,408 ±7.09 1,406 ±6 4.0% 2026-10-02 thinking

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 1,376 [1,365–1,386] Percentile 56 · #124 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/coding 1,417 ±6.88 1,397 ±6 89.2% 2026-10-02 thinking

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 Math 1,382 [1,367–1,398] Percentile 59 · #106 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,421 ±9.38 1,413 ±9 85.5% 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

B Writing 1,411 [1,401–1,420] Percentile 69 · #84 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,401 ±7.54 1,430 ±8 59.8% 2026-10-02 thinking
LMArena text/longer_query 1,422 ±6.11 1,421 ±6 33.1% 2026-10-02 thinking

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 Language: 繁體中文 1,413 [1,400–1,427] Percentile 68 · #72 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,470 ±9.27 1,430 ±7 92.8% 2026-10-02 thinking

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 Language: Deutsch 1,391 [1,368–1,414] Percentile 52 · #70 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/german 1,411 ±15 1,411 ±13 90.2% 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

B Language: English 1,408 [1,399–1,416] Percentile 66 · #92 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,432 ±4.59 1,422 ±5 93.5% 2026-10-02 thinking

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: Español 1,402 [1,377–1,427] Percentile 48 · #78 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/spanish 1,429 ±15 1,425 ±14 89.4% 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

C Language: 日本語 1,388 [1,367–1,409] Percentile 44 · #72 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/japanese 1,374 ±16 1,407 ±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: 한국어 1,370 [1,348–1,392] Percentile 36 · #88 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/korean 1,337 ±16 1,387 ±12 90.6% 2026-10-02 thinking

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: Polski 1,400 [1,386–1,415] Percentile 44 · #74 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/polish 1,411 ±9.72 1,416 ±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

B Language: Русский 1,395 [1,380–1,411] Percentile 58 · #98 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,405 ±9.47 1,411 ±8 92.4% 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

History

fedi-index by month
The same in a table
Model 2026-10
DeepSeek V3.1 1,377 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: DeepSeek V3.1

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