Skip to content
fedi.software

MiniMax M2.5 in fedi-index

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

fedi-index · General

B 1,371 ±9

Percentile 53 · #138

People prefer / Solves tasks

1,357 / 1,392

Human votes vs benchmarks, Elo-eq

Value

+9

$0.525 per 1M tokens, median of 33 providers · vendor $0.525

On your hardware

  • GPU 12 GB #26 · UD-TQ1_0 · strong compression · ≈18 tok/s
  • GPU 24 GB #29 · UD-IQ2_XXS · strong compression · ≈17 tok/s
  • 128 GB unified memory #15 · Q3_K_M · ≈20 tok/s

Score decomposition

B General 1,371 [1,362–1,380] Percentile 53 · #138 · components: 7
Source Component Value Elo-eq Weight Measured Variant
Epoch AI eci 147 ±1.52 1,412 ±12 35.8% 2026-02-12 —
LMArena text/overall 1,359 ±1.98 1,359 ±2 16.5% 2026-10-02 —
LMArena text/hard_prompts 1,372 ±2.44 1,369 ±2 16.4% 2026-10-02 —
LMArena text/instruction_following 1,353 ±3.01 1,368 ±3 8.1% 2026-10-02 —
LMArena text/expert 1,378 ±5.16 1,377 ±4 7.8% 2026-10-02 —
LMArena text/longer_query 1,366 ±2.94 1,369 ±3 4.1% 2026-10-02 —
LMArena text/multi_turn 1,364 ±3.78 1,366 ±3 4.0% 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

B Coding 1,375 [1,373–1,378] Percentile 55 · #125 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena webdev/overall 1,386 ±4.23 1,417 ±1 59.0% 2026-10-01 —
LMArena text/coding outlier 1,380 ±3.28 1,364 ±3 28.6% 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

C Math 1,351 [1,341–1,361] Percentile 46 · #137 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,378 ±6.05 1,372 ±6 87.4% 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 Writing 1,352 [1,347–1,357] Percentile 49 · #138 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,331 ±4.00 1,358 ±4 61.7% 2026-10-02 —
LMArena text/longer_query 1,366 ±2.94 1,369 ±3 32.1% 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

C Language: 繁體中文 1,357 [1,348–1,366] Percentile 41 · #132 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,392 ±5.90 1,368 ±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

C Language: Deutsch 1,354 [1,335–1,373] Percentile 35 · #94 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/german 1,361 ±12 1,369 ±11 91.4% 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: English 1,357 [1,352–1,362] Percentile 48 · #141 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,380 ±2.70 1,367 ±3 93.9% 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 Language: Español 1,346 [1,331–1,360] Percentile 20 · #120 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/spanish 1,355 ±8.42 1,357 ±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: Français 1,341 [1,326–1,356] Percentile 18 · #111 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/french 1,362 ±9.24 1,352 ±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

D Language: 日本語 1,248 [1,223–1,274] Percentile 16 · #108 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/japanese 1,169 ±19 1,254 ±14 89.2% 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

D Language: 한국어 1,300 [1,283–1,316] Percentile 18 · #111 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/korean 1,232 ±12 1,308 ±9 92.0% 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 Language: Polski 1,340 [1,324–1,355] Percentile 15 · #111 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/polish 1,333 ±9.89 1,351 ±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

C Language: Русский 1,359 [1,352–1,367] Percentile 42 · #136 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,359 ±4.38 1,370 ±4 93.7% 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

History

fedi-index by month
The same in a table
Model 2026-10
MiniMax M2.5 1,371 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: MiniMax M2.5

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