Qwen3.5 397B A17B in fedi-index
The score in every slice, what it is made of, value, local run and history.
fedi-index · General
B 1,412 ±5
Percentile 75 · #74
People prefer / Solves tasks
1,425 / 1,396
Human votes vs benchmarks, Elo-eq
Value
+27
$1.35 per 1M tokens, median of 31 providers · vendor $1.35
On your hardware
- 128 GB unified memory #35 · UD-IQ1_M · strong compression · ≈26 tok/s
Score decomposition
B General 1,412 [1,407–1,418] Percentile 75 · #74 · components: 8
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| Epoch AI |
eci
|
147 ±0.87 | 1,411 ±7 | 38.6% | 2026-02-13 | — |
| LMArena |
text/overall
|
1,438 ±1.59 | 1,438 ±2 | 15.4% | 2026-10-02 | — |
| LMArena |
text/hard_prompts
|
1,448 ±1.90 | 1,438 ±2 | 15.3% | 2026-10-02 | — |
| LMArena |
text/instruction_following
|
1,424 ±2.30 | 1,435 ±2 | 7.6% | 2026-10-02 | — |
| LMArena |
text/expert
|
1,462 ±3.49 | 1,445 ±3 | 7.5% | 2026-10-02 | — |
| LMArena |
text/longer_query
|
1,442 ±2.24 | 1,440 ±2 | 3.8% | 2026-10-02 | — |
| LMArena |
text/multi_turn
|
1,447 ±2.96 | 1,442 ±3 | 3.8% | 2026-10-02 | — |
| Epoch AI |
gpqa_diamond
decay ×0.26 |
0.86 ±0.02 | 1,441 ±11 | 1.3% | 2026-08-07 | none |
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 1,408 [1,406–1,410] Percentile 72 · #78 · components: 2 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
webdev/overall
|
1,399 ±2.60 | 1,421 ±1 | 45.7% | 2026-10-01 | — |
| LMArena |
text/coding
|
1,464 ±2.47 | 1,439 ±2 | 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
D Agents 1,388 [1,380–1,396] Percentile 7 · #54 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| τ²-bench (Sierra) |
tau2/banking_knowledge
decay ×0.76 |
0.10 ±0.03 | 1,430 ±5 | 81.4% | 2026-02-27 | enabled |
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 1,405 [1,398–1,412] Percentile 74 · #66 · components: 2
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/math
|
1,455 ±4.53 | 1,446 ±4 | 51.4% | 2026-10-02 | — |
| Epoch AI |
frontiermath_t1_3
decay ×0.86 |
0.31 ±0.03 | 1,389 ±6 | 41.6% | 2026-08-30 | none |
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 Vision 1,428 [1,425–1,431] Percentile 64 · #41 · components: 3 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
vision/overall
|
1,264 ±2.74 | 1,442 ±2 | 62.9% | 2026-10-02 | — |
| LMArena |
vision/ocr
|
1,272 ±2.70 | 1,444 ±3 | 15.7% | 2026-10-02 | — |
| LMArena |
vision/diagram
|
1,278 ±3.83 | 1,444 ±3 | 15.4% | 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 Writing 1,419 [1,415–1,423] Percentile 75 · #67 · components: 2 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/creative_writing
|
1,401 ±2.93 | 1,430 ±3 | 62.2% | 2026-10-02 | — |
| LMArena |
text/longer_query
|
1,442 ±2.24 | 1,440 ±2 | 31.7% | 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
A Language: 繁體中文 1,438 [1,432–1,444] Percentile 82 · #41 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/chinese
|
1,500 ±4.17 | 1,453 ±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: Deutsch 1,426 [1,412–1,440] Percentile 76 · #36 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/german
|
1,448 ±8.65 | 1,443 ±8 | 92.8% | 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: English 1,420 [1,416–1,424] Percentile 74 · #73 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/english
|
1,444 ±2.10 | 1,434 ±2 | 94.0% | 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: Español 1,422 [1,412–1,433] Percentile 68 · #48 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/spanish
|
1,443 ±6.27 | 1,438 ±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: Français 1,427 [1,416–1,438] Percentile 70 · #41 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/french
|
1,462 ±6.59 | 1,444 ±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
B Language: 日本語 1,429 [1,415–1,443] Percentile 80 · #27 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/japanese
|
1,427 ±10 | 1,447 ±8 | 92.7% | 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
B Language: 한국어 1,406 [1,395–1,418] Percentile 58 · #57 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/korean
|
1,382 ±8.55 | 1,422 ±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
A Language: Polski 1,434 [1,422–1,446] Percentile 80 · #27 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/polish
|
1,452 ±7.62 | 1,451 ±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
B Language: Русский 1,418 [1,413–1,424] Percentile 72 · #66 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/russian
|
1,428 ±3.32 | 1,432 ±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
History
The same in a table
| Model | 2026-10 |
|---|---|
| Qwen3.5 397B A17B | 1,412 B |
Badge
Show the model's fedi-index on your site or in a README. The badge updates itself; it links to this page.
Data CC BY 4.0: keep the link to fedi.software.