Skip to content
fedi.software

gpt-oss-120b in fedi-index

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

fedi-index · General

C 1,344 ±8

Percentile 45 · #161

People prefer / Solves tasks

1,345 / 1,341

Human votes vs benchmarks, Elo-eq

Value

-2

$0.2625 per 1M tokens, median of 52 providers

On your hardware

  • GPU 12 GB #8 · MXFP4 · ≈19 tok/s
  • GPU 24 GB #12 · MXFP4 · ≈25 tok/s
  • 128 GB unified memory #17 · MXFP4 · ≈44 tok/s

Score decomposition

C General 1,344 [1,335–1,352] Percentile 45 · #161 · components: 8
Source Component Value Elo-eq Weight Measured Variant
Epoch AI eci 140 ±1.93 1,358 ±15 27.9% 2025-08-05 —
LMArena text/overall 1,365 ±2.23 1,365 ±2 18.4% 2026-10-02 —
LMArena text/hard_prompts 1,364 ±2.88 1,362 ±3 18.3% 2026-10-02 —
LMArena text/instruction_following 1,318 ±3.60 1,335 ±3 9.0% 2026-10-02 —
LMArena text/expert 1,356 ±8.15 1,358 ±7 7.9% 2026-10-02 —
LMArena text/longer_query 1,319 ±3.92 1,325 ±4 4.4% 2026-10-02 —
LMArena text/multi_turn 1,340 ±4.45 1,344 ±4 4.4% 2026-10-02 —
Epoch AI gpqa_diamond decay ×0.26 0.76 ±0.03 1,392 ±13 1.5% 2025-12-11 high

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 Coding 1,348 [1,341–1,354] Percentile 43 · #158 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/coding 1,380 ±3.95 1,364 ±4 90.1% 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 Math 1,359 [1,347–1,370] Percentile 48 · #134 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,388 ±7.07 1,382 ±7 86.9% 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

C Writing 1,301 [1,295–1,308] Percentile 38 · #166 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,274 ±4.99 1,300 ±5 61.3% 2026-10-02 —
LMArena text/longer_query 1,319 ±3.92 1,325 ±4 32.3% 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 1,351 [1,342–1,360] Percentile 39 · #137 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,384 ±6.47 1,362 ±5 93.5% 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: Deutsch 1,349 [1,331–1,367] Percentile 34 · #95 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/german 1,354 ±12 1,362 ±10 91.7% 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

C Language: English 1,352 [1,346–1,357] Percentile 46 · #148 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,375 ±3.00 1,361 ±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

C Language: Español 1,372 [1,354–1,390] Percentile 30 · #105 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/spanish 1,388 ±11 1,388 ±10 91.7% 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: Français 1,343 [1,320–1,367] Percentile 22 · #106 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/french 1,370 ±14 1,359 ±13 90.0% 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

C Language: 日本語 1,362 [1,344–1,379] Percentile 33 · #86 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/japanese 1,332 ±13 1,376 ±10 91.8% 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

D Language: 한국어 1,335 [1,318–1,352] Percentile 23 · #105 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/korean 1,283 ±12 1,346 ±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 1,357 [1,345–1,368] Percentile 19 · #106 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/polish 1,354 ±7.40 1,368 ±6 93.2% 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

C Language: Русский 1,345 [1,335–1,356] Percentile 40 · #140 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,343 ±6.40 1,355 ±6 93.3% 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
gpt-oss-120b 1,344 C

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: gpt-oss-120b

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