Skip to content
fedi.software

Qwen3.5-35B-A3B in fedi-index

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

fedi-index · General

B 1,376 ±7

Percentile 56 · #131

People prefer / Solves tasks

1,383 / 1,365

Human votes vs benchmarks, Elo-eq

Value

+11

$0.5828 per 1M tokens, median of 20 providers · vendor $0.6875

On your hardware

  • GPU 12 GB #3 · Q8_0 · ≈18 tok/s
  • GPU 24 GB #7 · Q4_K_S · ≈191 tok/s
  • Mac 64 GB #5 · Q8_0 · ≈51 tok/s
  • 128 GB unified memory #7 · Q8_0 · ≈37 tok/s
  • CPU, 32 GB RAM #2 · Q4_K_M · ≈7 tok/s

Score decomposition

B General 1,376 [1,369–1,382] Percentile 56 · #131 · components: 8
Source Component Value Elo-eq Weight Measured Variant
Epoch AI eci 143 ±1.15 1,379 ±9 36.0% 2026-02-24 —
LMArena text/overall 1,394 ±2.15 1,394 ±2 16.1% 2026-10-02 —
LMArena text/hard_prompts 1,400 ±2.61 1,394 ±2 16.0% 2026-10-02 —
LMArena text/instruction_following 1,379 ±3.29 1,393 ±3 7.9% 2026-10-02 —
LMArena text/expert 1,407 ±5.74 1,400 ±5 7.5% 2026-10-02 —
LMArena text/longer_query 1,389 ±3.19 1,390 ±3 4.0% 2026-10-02 —
LMArena text/multi_turn 1,390 ±4.28 1,390 ±4 3.8% 2026-10-02 —
Epoch AI gpqa_diamond decay ×0.26 0.83 ±0.02 1,428 ±10 1.5% 2026-08-28 —

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 1,367 [1,363–1,371] Percentile 51 · #136 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena webdev/overall 1,253 ±8.88 1,380 ±2 45.5% 2026-10-01 —
LMArena text/coding 1,410 ±3.58 1,390 ±3 44.8% 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 Math 1,371 [1,360–1,383] Percentile 54 · #119 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,403 ±6.99 1,396 ±7 87.0% 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 1,369 [1,363–1,374] Percentile 54 · #124 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,347 ±4.58 1,374 ±5 61.3% 2026-10-02 —
LMArena text/longer_query 1,389 ±3.19 1,390 ±3 32.4% 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

B Language: 繁體中文 1,405 [1,395–1,415] Percentile 60 · #89 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,457 ±6.67 1,419 ±5 93.4% 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: Deutsch 1,358 [1,339–1,378] Percentile 36 · #92 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/german 1,367 ±13 1,374 ±11 91.1% 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

B Language: English 1,383 [1,378–1,389] Percentile 57 · #117 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,407 ±2.94 1,395 ±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 1,376 [1,359–1,392] Percentile 31 · #103 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/spanish 1,391 ±9.80 1,390 ±9 92.1% 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: Français 1,382 [1,364–1,400] Percentile 36 · #87 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/french 1,412 ±11 1,398 ±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

C Language: 한국어 1,384 [1,365–1,403] Percentile 42 · #80 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/korean 1,355 ±14 1,401 ±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: Polski 1,379 [1,361–1,397] Percentile 32 · #89 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/polish 1,385 ±12 1,395 ±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

C Language: Русский 1,374 [1,365–1,382] Percentile 47 · #124 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,376 ±5.10 1,386 ±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

History

fedi-index by month
The same in a table
Model 2026-10
Qwen3.5-35B-A3B 1,376 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: Qwen3.5-35B-A3B

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