GPT-5.4 in fedi-index
The score in every slice, what it is made of, value, local run and history.
fedi-index · General
A 1,459 ±5
Percentile 94 · #20
People prefer / Solves tasks
1,457 / 1,462
Human votes vs benchmarks, Elo-eq
Value
+39
$5.62 per 1M tokens, median of 35 providers · vendor $5.62
On your hardware
Closed weights: cloud only.
Score decomposition
A General 1,459 [1,454–1,464] Percentile 94 · #20 · components: 9
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| Epoch AI |
eci
|
157 ±1.01 | 1,492 ±8 | 33.0% | 2026-03-05 | — |
| LMArena |
text/overall
|
1,469 ±1.91 | 1,469 ±2 | 15.8% | 2026-10-02 | high |
| LMArena |
text/hard_prompts
|
1,485 ±2.29 | 1,472 ±2 | 15.8% | 2026-10-02 | high |
| LMArena |
text/instruction_following
|
1,469 ±2.78 | 1,479 ±3 | 7.8% | 2026-10-02 | high |
| LMArena |
text/expert
|
1,507 ±4.23 | 1,482 ±4 | 7.6% | 2026-10-02 | high |
| LMArena |
text/longer_query
|
1,474 ±2.68 | 1,469 ±2 | 3.9% | 2026-10-02 | high |
| LMArena |
text/multi_turn
|
1,482 ±3.36 | 1,473 ±3 | 3.9% | 2026-10-02 | high |
| Epoch AI |
simpleqa_verified
decay ×0.61 |
0.45 ±0.02 | 1,439 ±5 | 3.9% | 2026-08-27 | xhigh |
| Epoch AI |
gpqa_diamond
decay ×0.26 |
0.93 ±0.02 | 1,474 ±8 | 1.4% | 2026-03-06 | xhigh |
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 Coding 1,434 [1,431–1,438] Percentile 86 · #40 · components: 3
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/coding
|
1,497 ±2.96 | 1,467 ±3 | 38.9% | 2026-10-02 | high |
| LMArena |
webdev/overall
|
1,465 ±9.63 | 1,440 ±3 | 38.8% | 2026-10-01 | high-codex-harness |
| Epoch AI |
swe_bench_verified
decay ×0.37 |
0.77 ±0.02 | 1,468 ±10 | 13.9% | 2026-03-06 | 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
B Agents 1,447 [1,442–1,452] Percentile 65 · #21 · components: 2
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
agent/overall
|
0.01 ±0.00 | 1,461 ±2 | 66.5% | 2026-10-02 | high |
| τ²-bench (Sierra) |
tau2/banking_knowledge
decay ×0.76 |
0.39 ±0.05 | 1,479 ±8 | 26.4% | 2026-05-06 | xhigh |
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 Math 1,461 [1,454–1,467] Percentile 95 · #14 · components: 3
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/math
|
1,488 ±5.40 | 1,478 ±5 | 52.6% | 2026-10-02 | high |
| Epoch AI |
frontiermath_t1_3
decay ×0.86 |
0.79 ±0.02 | 1,491 ±5 | 30.2% | 2026-06-11 | xhigh |
| Epoch AI |
frontiermath_t4
decay ×0.86 |
0.49 ±0.08 | 1,471 ±13 | 9.8% | 2026-06-11 | xhigh |
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 Vision 1,462 [1,459–1,466] Percentile 91 · #11 · components: 3 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
vision/overall
|
1,302 ±2.98 | 1,477 ±3 | 62.9% | 2026-10-02 | high |
| LMArena |
vision/ocr
|
1,311 ±2.95 | 1,482 ±3 | 15.7% | 2026-10-02 | high |
| LMArena |
vision/diagram
|
1,326 ±4.22 | 1,487 ±4 | 15.3% | 2026-10-02 | 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
A Writing 1,452 [1,448–1,457] Percentile 87 · #35 · components: 2 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/creative_writing
|
1,439 ±3.51 | 1,468 ±4 | 62.0% | 2026-10-02 | high |
| LMArena |
text/longer_query
|
1,474 ±2.68 | 1,469 ±2 | 31.9% | 2026-10-02 | 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
A Language: 繁體中文 1,452 [1,445–1,460] Percentile 90 · #23 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/chinese
|
1,520 ±5.07 | 1,469 ±4 | 93.7% | 2026-10-02 | 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
A Language: Deutsch 1,445 [1,430–1,460] Percentile 90 · #15 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/german
|
1,472 ±9.53 | 1,464 ±8 | 92.5% | 2026-10-02 | 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
A Language: English 1,443 [1,438–1,448] Percentile 88 · #35 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/english
|
1,467 ±2.51 | 1,458 ±3 | 94.0% | 2026-10-02 | 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
B Language: Español 1,431 [1,419–1,444] Percentile 77 · #35 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/spanish
|
1,454 ±7.25 | 1,448 ±7 | 93.0% | 2026-10-02 | 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
A Language: Français 1,453 [1,441–1,465] Percentile 91 · #13 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/french
|
1,493 ±7.17 | 1,472 ±7 | 93.1% | 2026-10-02 | 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
A Language: 日本語 1,466 [1,449–1,483] Percentile 92 · #11 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/japanese
|
1,483 ±13 | 1,489 ±10 | 92.0% | 2026-10-02 | 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
A Language: 한국어 1,453 [1,440–1,467] Percentile 91 · #13 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/korean
|
1,450 ±9.98 | 1,473 ±8 | 92.7% | 2026-10-02 | 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
A Language: Polski 1,449 [1,436–1,462] Percentile 89 · #15 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/polish
|
1,471 ±8.48 | 1,468 ±7 | 92.9% | 2026-10-02 | 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
A Language: Русский 1,462 [1,455–1,469] Percentile 93 · #17 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/russian
|
1,480 ±3.98 | 1,479 ±4 | 93.8% | 2026-10-02 | 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
History
The same in a table
| Model | 2026-10 |
|---|---|
| GPT-5.4 | 1,459 A |
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.