Gemma 4 26B A4B in fedi-index
The score in every slice, what it is made of, value, local run and history.
fedi-index · General
B 1,391 ±8
Percentile 63 · #108
People prefer / Solves tasks
1,416 / 1,358
Human votes vs benchmarks, Elo-eq
Value
+54
$0.1938 per 1M tokens, median of 22 providers
On your hardware
- GPU 12 GB #2 · UD-Q8_K_XL · ≈15 tok/s
- GPU 24 GB #3 · UD-Q5_K_S · ≈129 tok/s
- Mac 64 GB #2 · UD-Q8_K_XL · ≈40 tok/s
- 128 GB unified memory #4 · UD-Q8_K_XL · ≈29 tok/s
- CPU, 32 GB RAM #1 · UD-Q4_K_XL · ≈6 tok/s
Score decomposition
B General 1,391 [1,384–1,399] Percentile 63 · #108 · components: 8
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| Epoch AI |
eci
|
142 ±1.20 | 1,373 ±10 | 37.8% | 2026-04-02 | — |
| LMArena |
text/overall
|
1,434 ±3.79 | 1,434 ±4 | 16.6% | 2026-10-02 | — |
| LMArena |
text/hard_prompts
|
1,438 ±4.97 | 1,429 ±4 | 16.2% | 2026-10-02 | — |
| LMArena |
text/instruction_following
|
1,420 ±6.86 | 1,432 ±6 | 7.4% | 2026-10-02 | — |
| LMArena |
text/expert
|
1,447 ±13 | 1,433 ±11 | 5.9% | 2026-10-02 | — |
| LMArena |
text/longer_query
|
1,427 ±7.00 | 1,426 ±6 | 3.7% | 2026-10-02 | — |
| LMArena |
text/multi_turn
|
1,441 ±8.62 | 1,436 ±8 | 3.5% | 2026-10-02 | — |
| Epoch AI |
gpqa_diamond
decay ×0.26 |
0.73 ±0.03 | 1,380 ±15 | 1.2% | 2026-08-06 | minimal |
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 1,394 [1,388–1,399] Percentile 65 · #97 · components: 2 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
webdev/overall
|
1,358 ±8.63 | 1,409 ±2 | 48.3% | 2026-10-01 | — |
| LMArena |
text/coding
|
1,446 ±7.49 | 1,423 ±7 | 41.4% | 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 Math 1,412 [1,390–1,433] Percentile 78 · #56 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/math
|
1,469 ±14 | 1,459 ±14 | 81.6% | 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 Vision 1,423 [1,419–1,427] Percentile 60 · #46 · components: 3 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
vision/overall
|
1,260 ±3.30 | 1,438 ±3 | 62.9% | 2026-10-02 | — |
| LMArena |
vision/ocr
|
1,268 ±3.32 | 1,440 ±3 | 15.6% | 2026-10-02 | — |
| LMArena |
vision/diagram
|
1,264 ±4.48 | 1,431 ±4 | 15.3% | 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 Writing 1,411 [1,399–1,423] Percentile 69 · #83 · components: 2 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/creative_writing
|
1,401 ±9.29 | 1,430 ±10 | 58.2% | 2026-10-02 | — |
| LMArena |
text/longer_query
|
1,427 ±7.00 | 1,426 ±6 | 34.2% | 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: 繁體中文 1,429 [1,410–1,447] Percentile 77 · #53 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/chinese
|
1,495 ±13 | 1,449 ±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
B Language: English 1,418 [1,406–1,430] Percentile 72 · #77 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/english
|
1,444 ±6.09 | 1,434 ±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 1,420 [1,405–1,435] Percentile 73 · #63 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/russian
|
1,434 ±9.20 | 1,437 ±8 | 92.5% | 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
History
The same in a table
| Model | 2026-10 |
|---|---|
| Gemma 4 26B A4B | 1,391 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.