Skip to content
fedi.software

Ministral 3 8B 2512 in fedi-index

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

fedi-index · General

D 1,199 ±5

Percentile 23 · #226 · provisional

People prefer / Solves tasks

1,198 / —

Human votes vs benchmarks, Elo-eq

Value

-133

$0.15 per 1M tokens, median of 7 providers · vendor $0.10

On your hardware

  • GPU 12 GB #22 · UD-Q6_K_XL · ≈35 tok/s
  • GPU 24 GB #23 · Q8_0 · ≈78 tok/s
  • Mac 64 GB #18 · Q8_0 · ≈40 tok/s
  • 128 GB unified memory #33 · Q8_0 · ≈19 tok/s
  • CPU, 32 GB RAM #15 · Q4_K_M · ≈5 tok/s

Score decomposition

D General 1,199 [1,194–1,204] Percentile 23 · #226 · components: 6 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/overall 1,191 ±4.58 1,191 ±5 28.0% 2026-10-02 —
LMArena text/hard_prompts 1,191 ±7.68 1,206 ±7 25.2% 2026-10-02 —
LMArena text/instruction_following 1,161 ±6.48 1,186 ±6 13.1% 2026-10-02 —
LMArena text/expert 1,170 ±15 1,206 ±12 9.1% 2026-10-02 —
LMArena text/multi_turn 1,166 ±9.62 1,187 ±9 5.7% 2026-10-02 —
LMArena text/longer_query 1,212 ±10 1,225 ±10 5.4% 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 Coding 1,204 [1,190–1,219] Percentile 22 · #217 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/coding 1,202 ±9.46 1,205 ±8 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 Math 1,192 [1,176–1,208] Percentile 23 · #196 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,188 ±10 1,190 ±10 85.0% 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 Writing 1,207 [1,193–1,222] Percentile 24 · #203 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,175 ±11 1,198 ±11 58.8% 2026-10-02 —
LMArena text/longer_query 1,212 ±10 1,225 ±10 32.6% 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: 繁體中文 1,209 [1,191–1,227] Percentile 17 · #186 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,193 ±13 1,210 ±10 91.6% 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: English 1,191 [1,179–1,202] Percentile 24 · #209 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,212 ±6.07 1,190 ±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: Русский 1,221 [1,205–1,238] Percentile 17 · #195 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,195 ±10 1,223 ±9 92.2% 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

History

fedi-index by month
The same in a table
Model 2026-10
Ministral 3 8B 2512 1,199 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: Ministral 3 8B 2512

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