DeepSeek V4 Flash 0731 in fedi-index
The score in every slice, what it is made of, value, local run and history.
fedi-index · General
A 1,424 ±13
Percentile 80 · #58 · provisional
People prefer / Solves tasks
— / 1,445
Human votes vs benchmarks, Elo-eq
Value
+89
$0.175 per 1M tokens, median of 45 providers
On your hardware
- 128 GB unified memory #10 · UD-IQ3_XXS · ≈11 tok/s
Score decomposition
A General 1,424 [1,412–1,437] Percentile 80 · #58 · components: 3 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| Epoch AI |
eci
|
154 ±1.09 | 1,473 ±9 | 75.3% | 2026-07-31 | — |
| Epoch AI |
simpleqa_verified
outlier decay ×0.61 |
0.34 ±0.01 | 1,403 ±5 | 4.6% | 2026-08-27 | max |
| Epoch AI |
gpqa_diamond
decay ×0.26 |
0.91 ±0.02 | 1,463 ±8 | 3.4% | 2026-08-02 | max |
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 1,404 [1,395–1,414] Percentile 74 · #68 · components: 2 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| Epoch AI |
frontiermath_t1_3
decay ×0.86 |
0.58 ±0.03 | 1,446 ±6 | 60.8% | 2026-08-02 | max |
| Epoch AI |
frontiermath_t4
decay ×0.86 |
0.24 ±0.07 | 1,432 ±11 | 23.6% | 2026-08-02 | max |
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 |
|---|---|
| DeepSeek V4 Flash 0731 | 1,424 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.