Skip to content
fedi.software

Qwen3 Next 80B A3B Instruct in fedi-index

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

fedi-index · General

B 1,388 ±3

Percentile 62 · #113 · provisional

People prefer / Solves tasks

1,400 / —

Human votes vs benchmarks, Elo-eq

Value

+32

$0.4125 per 1M tokens, median of 18 providers · vendor $0.875

On your hardware

  • GPU 12 GB #1 · Q5_K_M · ≈24 tok/s
  • GPU 24 GB #4 · UD-Q6_K_XL · ≈26 tok/s
  • Mac 64 GB #3 · Q4_K_S · ≈84 tok/s
  • 128 GB unified memory #5 · Q8_0 · ≈36 tok/s
  • CPU, 32 GB RAM #16 · UD-IQ2_XXS · strong compression · ≈7 tok/s

Score decomposition

B General 1,388 [1,386–1,391] Percentile 62 · #113 · components: 6 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/overall 1,417 ±2.42 1,417 ±2 26.2% 2026-10-02 —
LMArena text/hard_prompts 1,428 ±3.06 1,420 ±3 26.0% 2026-10-02 —
LMArena text/instruction_following 1,388 ±3.85 1,402 ±4 12.7% 2026-10-02 —
LMArena text/expert 1,416 ±9.04 1,408 ±7 10.8% 2026-10-02 —
LMArena text/longer_query 1,402 ±4.20 1,402 ±4 6.3% 2026-10-02 —
LMArena text/multi_turn 1,415 ±4.81 1,413 ±4 6.2% 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 Coding 1,396 [1,389–1,403] Percentile 66 · #94 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/coding 1,440 ±4.39 1,417 ±4 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

B Math 1,399 [1,385–1,413] Percentile 69 · #79 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,439 ±8.56 1,431 ±8 86.0% 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 Writing 1,364 [1,356–1,371] Percentile 52 · #129 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,333 ±5.54 1,360 ±6 60.9% 2026-10-02 —
LMArena text/longer_query 1,402 ±4.20 1,402 ±4 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

B Language: 繁體中文 1,408 [1,398–1,419] Percentile 63 · #84 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,462 ±7.33 1,423 ±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

B Language: Deutsch 1,397 [1,376–1,418] Percentile 54 · #66 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/german 1,416 ±13 1,416 ±12 91.0% 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 Language: English 1,398 [1,392–1,404] Percentile 62 · #103 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,422 ±3.27 1,411 ±3 93.8% 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

B Language: Español 1,413 [1,393–1,432] Percentile 57 · #64 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/spanish 1,437 ±12 1,433 ±11 91.3% 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: Français 1,379 [1,354–1,404] Percentile 33 · #91 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/french 1,415 ±16 1,400 ±14 89.3% 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

B Language: 日本語 1,404 [1,381–1,426] Percentile 56 · #57 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/japanese 1,398 ±17 1,425 ±13 90.3% 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: 한국어 1,391 [1,372–1,410] Percentile 47 · #72 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/korean 1,365 ±14 1,408 ±10 91.5% 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: Polski 1,391 [1,376–1,405] Percentile 38 · #81 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/polish 1,398 ±9.44 1,406 ±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

B Language: Русский 1,396 [1,384–1,408] Percentile 59 · #97 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,404 ±7.33 1,410 ±7 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

History

fedi-index by month
The same in a table
Model 2026-10
Qwen3 Next 80B A3B Instruct 1,388 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: Qwen3 Next 80B A3B Instruct

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