Kimi K3 in fedi-index
The score in every slice, what it is made of, value, local run and history.
fedi-index · General
S 1,466 ±6
Percentile 96 · #14
People prefer / Solves tasks
1,466 / 1,465
Human votes vs benchmarks, Elo-eq
Value
+44
$6 per 1M tokens, median of 62 providers · vendor $6
On your hardware
Does not fit the reference setups.
Score decomposition
S General 1,466 [1,459–1,472] Percentile 96 · #14 · components: 9
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| Epoch AI |
eci
|
157 ±1.24 | 1,497 ±10 | 30.8% | 2026-07-16 | — |
| LMArena |
text/overall
|
1,476 ±2.43 | 1,476 ±2 | 16.4% | 2026-10-02 | max |
| LMArena |
text/hard_prompts
|
1,496 ±2.81 | 1,482 ±2 | 16.3% | 2026-10-02 | max |
| LMArena |
text/instruction_following
|
1,483 ±3.52 | 1,492 ±3 | 8.0% | 2026-10-02 | max |
| LMArena |
text/expert
|
1,521 ±6.03 | 1,494 ±5 | 7.6% | 2026-10-02 | max |
| LMArena |
text/longer_query
|
1,494 ±3.31 | 1,488 ±3 | 4.0% | 2026-10-02 | max |
| Epoch AI |
simpleqa_verified
decay ×0.61 |
0.51 ±0.02 | 1,456 ±5 | 4.0% | 2026-08-27 | max |
| LMArena |
text/multi_turn
|
1,488 ±4.83 | 1,478 ±4 | 3.9% | 2026-10-02 | max |
| Epoch AI |
gpqa_diamond
decay ×0.26 |
0.93 ±0.01 | 1,473 ±7 | 1.6% | 2026-07-16 | 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
A Coding 1,460 [1,456–1,463] Percentile 96 · #12 · components: 2 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
webdev/overall
|
1,658 ±3.25 | 1,494 ±1 | 46.2% | 2026-10-01 | max |
| LMArena |
text/coding
|
1,510 ±3.88 | 1,479 ±4 | 44.1% | 2026-10-02 | 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 Agents 1,456 [1,452–1,461] Percentile 79 · #13 · components: 2
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
agent/overall
|
0.04 ±0.00 | 1,476 ±1 | 66.8% | 2026-10-02 | max |
| τ²-bench (Sierra) |
tau2/banking_knowledge
decay ×0.76 |
0.37 ±0.05 | 1,476 ±8 | 26.2% | 2026-07-24 | 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
A Math 1,456 [1,447–1,465] Percentile 93 · #19 · components: 3
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/math
|
1,494 ±8.66 | 1,484 ±8 | 48.9% | 2026-10-02 | max |
| Epoch AI |
frontiermath_t1_3
decay ×0.86 |
0.72 ±0.03 | 1,477 ±6 | 32.1% | 2026-07-17 | max |
| Epoch AI |
frontiermath_t4
decay ×0.86 |
0.39 ±0.08 | 1,455 ±12 | 11.0% | 2026-07-17 | 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
A Writing 1,468 [1,462–1,473] Percentile 93 · #19 · components: 2 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/creative_writing
|
1,454 ±4.45 | 1,484 ±4 | 61.5% | 2026-10-02 | max |
| LMArena |
text/longer_query
|
1,494 ±3.31 | 1,488 ±3 | 32.2% | 2026-10-02 | 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
A Language: 繁體中文 1,458 [1,448–1,468] Percentile 94 · #15 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/chinese
|
1,528 ±6.88 | 1,476 ±6 | 93.4% | 2026-10-02 | 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
A Language: Deutsch 1,445 [1,425–1,466] Percentile 91 · #14 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/german
|
1,478 ±13 | 1,469 ±11 | 91.0% | 2026-10-02 | 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
A Language: English 1,458 [1,452–1,464] Percentile 95 · #15 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/english
|
1,482 ±3.28 | 1,475 ±4 | 93.8% | 2026-10-02 | 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
A Language: Español 1,448 [1,431–1,465] Percentile 90 · #15 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/spanish
|
1,477 ±10 | 1,470 ±10 | 91.9% | 2026-10-02 | 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
A Language: Français 1,448 [1,432–1,465] Percentile 88 · #17 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/french
|
1,491 ±10 | 1,470 ±9 | 92.0% | 2026-10-02 | 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
A Language: 日本語 1,468 [1,448–1,487] Percentile 94 · #9 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/japanese
|
1,488 ±14 | 1,493 ±11 | 91.3% | 2026-10-02 | 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
A Language: 한국어 1,459 [1,442–1,476] Percentile 93 · #11 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/korean
|
1,460 ±12 | 1,481 ±9 | 92.0% | 2026-10-02 | 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
A Language: Polski 1,444 [1,423–1,464] Percentile 86 · #19 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/polish
|
1,471 ±14 | 1,468 ±12 | 90.9% | 2026-10-02 | 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
A Language: Русский 1,463 [1,454–1,472] Percentile 94 · #16 · components: 1 provisional
| Source | Component | Value | Elo-eq | Weight | Measured | Variant |
|---|---|---|---|---|---|---|
| LMArena |
text/russian
|
1,482 ±5.34 | 1,481 ±5 | 93.6% | 2026-10-02 | 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 |
|---|---|
| Kimi K3 | 1,466 S |
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.