Skip to content

docs(go): update Kimi K3 request estimates #37725

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
fwang merged 5 commits into from
Jul 19, 2026
Merged

docs(go): update Kimi K3 request estimates #37725

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
e2e2a37
docs(go): use Kimi K3 p50 estimates
fwang Jul 19, 2026
a61555a
docs(go): round Kimi K3 token profile
fwang Jul 19, 2026
7587c47
docs(go): adjust Kimi K3 five-hour estimate
fwang Jul 19, 2026
f3fb500
docs(go): translate Kimi K3 estimates
fwang Jul 19, 2026
98b539b
docs(go): remove P50 label
fwang Jul 19, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion packages/console/app/src/routes/go/index.tsx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -64,7 +64,7 @@ function LimitsGraph(props: { href: string }) {
const baseline = 100
const graph = [
{ id: "grok-4.5", name: "Grok 4.5", req: 120, d: "50ms" },
{ id: "kimi-k3", name: "Kimi K3 (2x usage)", req: 280, baseReq: 140, d: "75ms" },
{ id: "kimi-k3", name: "Kimi K3 (2x usage)", req: 220, baseReq: 110, d: "75ms" },
{ id: "glm-5.2", name: "GLM-5.2", req: 880, d: "100ms" },
{ id: "qwen3.7-max", name: "Qwen3.7 Max", req: 950, d: "110ms" },
{ id: "kimi-k2.7-code", name: "Kimi K2.7 Code", req: 1150, d: "150ms" },
Expand Down
7 changes: 4 additions & 3 deletions packages/web/src/content/docs/ar/go.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,7 +86,7 @@ OpenCode Go هو اشتراك منخفض التكلفة — **$5 للشهر ال
| Grok 4.5 | 120 | 300 | 600 |
| GLM-5.2 | 880 | 2,150 | 4,300 |
| GLM-5.1 | 880 | 2,150 | 4,300 |
| Kimi K3 | 140 | 340 | 680 |
| Kimi K3 | 110 | 250 | 490 |
| Kimi K2.7 Code | 1,350 | 4,630 | 9,250 |
| Kimi K2.6 | 1,150 | 2,880 | 5,750 |
| MiMo-V2.5 | 30,100 | 75,200 | 150,400 |
Expand All @@ -99,11 +99,12 @@ OpenCode Go هو اشتراك منخفض التكلفة — **$5 للشهر ال
| DeepSeek V4 Pro | 3,450 | 8,550 | 17,150 |
| DeepSeek V4 Flash | 31,650 | 79,050 | 158,150 |

تستند التقديرات إلى متوسط أنماط الطلبات المرصودة:
تستند التقديرات إلى أنماط الطلبات المرصودة:

- Grok 4.5 — ‏1,100 input، و71,500 cached، و220 output tokens لكل طلب
- GLM-5.2/5.1 — ‏700 input، و52,000 cached، و150 output tokens لكل طلب
- Kimi K3/K2.7/K2.6 — ‏870 input، و55,000 cached، و200 output tokens لكل طلب
- Kimi K3 — ‏1,050 input، و76,500 cached، و300 output tokens لكل طلب
- Kimi K2.7/K2.6 — ‏870 input، و55,000 cached، و200 output tokens لكل طلب
- DeepSeek V4 Pro — ‏750 input، و82,000 cached، و290 output tokens لكل طلب
- DeepSeek V4 Flash — ‏790 input، و68,000 cached، و280 output tokens لكل طلب
- MiniMax M3 — ‏510 input، و56,000 cached، و190 output tokens لكل طلب
Expand Down
7 changes: 4 additions & 3 deletions packages/web/src/content/docs/bs/go.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -96,7 +96,7 @@ Tabela ispod pruža procijenjeni broj zahtjeva na osnovu tipičnih obrazaca kori
| Grok 4.5 | 120 | 300 | 600 |
| GLM-5.2 | 880 | 2,150 | 4,300 |
| GLM-5.1 | 880 | 2,150 | 4,300 |
| Kimi K3 | 140 | 340 | 680 |
| Kimi K3 | 110 | 250 | 490 |
| Kimi K2.7 Code | 1,350 | 4,630 | 9,250 |
| Kimi K2.6 | 1,150 | 2,880 | 5,750 |
| MiMo-V2.5 | 30,100 | 75,200 | 150,400 |
Expand All @@ -109,11 +109,12 @@ Tabela ispod pruža procijenjeni broj zahtjeva na osnovu tipičnih obrazaca kori
| DeepSeek V4 Pro | 3,450 | 8,550 | 17,150 |
| DeepSeek V4 Flash | 31,650 | 79,050 | 158,150 |

Procjene se zasnivaju na zapaženim prosječnim obrascima zahtjeva:
Procjene se zasnivaju na zapaženim obrascima zahtjeva:

- Grok 4.5 — 1,100 ulaznih, 71,500 keširanih, 220 izlaznih tokena po zahtjevu
- GLM-5.2/5.1 — 700 ulaznih (input), 52,000 keširanih, 150 izlaznih (output) tokena po zahtjevu
- Kimi K3/K2.7/K2.6 — 870 ulaznih, 55,000 keširanih, 200 izlaznih tokena po zahtjevu
- Kimi K3 — 1,050 ulaznih, 76,500 keširanih, 300 izlaznih tokena po zahtjevu
- Kimi K2.7/K2.6 — 870 ulaznih, 55,000 keširanih, 200 izlaznih tokena po zahtjevu
- DeepSeek V4 Pro — 750 ulaznih, 82,000 keširanih, 290 izlaznih tokena po zahtjevu
- DeepSeek V4 Flash — 790 ulaznih, 68,000 keširanih, 280 izlaznih tokena po zahtjevu
- MiniMax M3 — 510 ulaznih, 56,000 keširanih, 190 izlaznih tokena po zahtjevu
Expand Down
7 changes: 4 additions & 3 deletions packages/web/src/content/docs/da/go.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -96,7 +96,7 @@ Tabellen nedenfor giver et estimeret antal anmodninger baseret på typiske Go-fo
| Grok 4.5 | 120 | 300 | 600 |
| GLM-5.2 | 880 | 2,150 | 4,300 |
| GLM-5.1 | 880 | 2,150 | 4,300 |
| Kimi K3 | 140 | 340 | 680 |
| Kimi K3 | 110 | 250 | 490 |
| Kimi K2.7 Code | 1,350 | 4,630 | 9,250 |
| Kimi K2.6 | 1,150 | 2,880 | 5,750 |
| MiMo-V2.5 | 30,100 | 75,200 | 150,400 |
Expand All @@ -109,11 +109,12 @@ Tabellen nedenfor giver et estimeret antal anmodninger baseret på typiske Go-fo
| DeepSeek V4 Pro | 3,450 | 8,550 | 17,150 |
| DeepSeek V4 Flash | 31,650 | 79,050 | 158,150 |

Estimaterne er baseret på observerede gennemsnitlige anmodningsmønstre:
Estimaterne er baseret på observerede anmodningsmønstre:

- Grok 4.5 — 1.100 input, 71.500 cachelagrede, 220 output-tokens pr. anmodning
- GLM-5.2/5.1 — 700 input, 52.000 cachelagrede, 150 output-tokens pr. anmodning
- Kimi K3/K2.7/K2.6 — 870 input, 55.000 cachelagrede, 200 output-tokens pr. anmodning
- Kimi K3 — 1.050 input, 76.500 cachelagrede, 300 output-tokens pr. anmodning
- Kimi K2.7/K2.6 — 870 input, 55.000 cachelagrede, 200 output-tokens pr. anmodning
- DeepSeek V4 Pro — 750 input, 82.000 cachelagrede, 290 output-tokens pr. anmodning
- DeepSeek V4 Flash — 790 input, 68.000 cachelagrede, 280 output-tokens pr. anmodning
- MiniMax M3 — 510 input, 56.000 cachelagrede, 190 output-tokens pr. anmodning
Expand Down
7 changes: 4 additions & 3 deletions packages/web/src/content/docs/de/go.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -88,7 +88,7 @@ Die folgende Tabelle zeigt eine geschätzte Anzahl von Anfragen basierend auf ty
| Grok 4.5 | 120 | 300 | 600 |
| GLM-5.2 | 880 | 2,150 | 4,300 |
| GLM-5.1 | 880 | 2,150 | 4,300 |
| Kimi K3 | 140 | 340 | 680 |
| Kimi K3 | 110 | 250 | 490 |
| Kimi K2.7 Code | 1,350 | 4,630 | 9,250 |
| Kimi K2.6 | 1,150 | 2,880 | 5,750 |
| MiMo-V2.5 | 30,100 | 75,200 | 150,400 |
Expand All @@ -101,11 +101,12 @@ Die folgende Tabelle zeigt eine geschätzte Anzahl von Anfragen basierend auf ty
| DeepSeek V4 Pro | 3,450 | 8,550 | 17,150 |
| DeepSeek V4 Flash | 31,650 | 79,050 | 158,150 |

Die Schätzungen basieren auf beobachteten durchschnittlichen Anfragemustern:
Die Schätzungen basieren auf beobachteten Anfragemustern:

- Grok 4.5 — 1.100 Input-, 71.500 Cached-, 220 Output-Tokens pro Anfrage
- GLM-5.2/5.1 — 700 Input-, 52.000 Cached-, 150 Output-Tokens pro Anfrage
- Kimi K3/K2.7/K2.6 — 870 Input-, 55.000 Cached-, 200 Output-Tokens pro Anfrage
- Kimi K3 — 1.050 Input-, 76.500 Cached-, 300 Output-Tokens pro Anfrage
- Kimi K2.7/K2.6 — 870 Input-, 55.000 Cached-, 200 Output-Tokens pro Anfrage
- DeepSeek V4 Pro — 750 Input-, 82.000 Cached-, 290 Output-Tokens pro Anfrage
- DeepSeek V4 Flash — 790 Input-, 68.000 Cached-, 280 Output-Tokens pro Anfrage
- MiniMax M3 — 510 Input-, 56.000 Cached-, 190 Output-Tokens pro Anfrage
Expand Down
7 changes: 4 additions & 3 deletions packages/web/src/content/docs/es/go.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -96,7 +96,7 @@ La siguiente tabla proporciona una cantidad estimada de peticiones basada en los
| Grok 4.5 | 120 | 300 | 600 |
| GLM-5.2 | 880 | 2,150 | 4,300 |
| GLM-5.1 | 880 | 2,150 | 4,300 |
| Kimi K3 | 140 | 340 | 680 |
| Kimi K3 | 110 | 250 | 490 |
| Kimi K2.7 Code | 1,350 | 4,630 | 9,250 |
| Kimi K2.6 | 1,150 | 2,880 | 5,750 |
| MiMo-V2.5 | 30,100 | 75,200 | 150,400 |
Expand All @@ -109,11 +109,12 @@ La siguiente tabla proporciona una cantidad estimada de peticiones basada en los
| DeepSeek V4 Pro | 3,450 | 8,550 | 17,150 |
| DeepSeek V4 Flash | 31,650 | 79,050 | 158,150 |

Las estimaciones se basan en los patrones de peticiones promedio observados:
Las estimaciones se basan en los patrones de peticiones observados:

- Grok 4.5 — 1,100 tokens de entrada, 71,500 en caché, 220 tokens de salida por petición
- GLM-5.2/5.1 — 700 tokens de entrada, 52,000 en caché, 150 tokens de salida por petición
- Kimi K3/K2.7/K2.6 — 870 tokens de entrada, 55,000 en caché, 200 tokens de salida por petición
- Kimi K3 — 1,050 tokens de entrada, 76,500 en caché, 300 tokens de salida por petición
- Kimi K2.7/K2.6 — 870 tokens de entrada, 55,000 en caché, 200 tokens de salida por petición
- DeepSeek V4 Pro — 750 tokens de entrada, 82,000 en caché, 290 tokens de salida por petición
- DeepSeek V4 Flash — 790 tokens de entrada, 68,000 en caché, 280 tokens de salida por petición
- MiniMax M3 — 510 tokens de entrada, 56,000 en caché, 190 tokens de salida por petición
Expand Down
7 changes: 4 additions & 3 deletions packages/web/src/content/docs/fr/go.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,7 +86,7 @@ Le tableau ci-dessous fournit une estimation du nombre de requêtes basée sur d
| Grok 4.5 | 120 | 300 | 600 |
| GLM-5.2 | 880 | 2,150 | 4,300 |
| GLM-5.1 | 880 | 2,150 | 4,300 |
| Kimi K3 | 140 | 340 | 680 |
| Kimi K3 | 110 | 250 | 490 |
| Kimi K2.7 Code | 1,350 | 4,630 | 9,250 |
| Kimi K2.6 | 1,150 | 2,880 | 5,750 |
| MiMo-V2.5 | 30,100 | 75,200 | 150,400 |
Expand All @@ -99,11 +99,12 @@ Le tableau ci-dessous fournit une estimation du nombre de requêtes basée sur d
| DeepSeek V4 Pro | 3,450 | 8,550 | 17,150 |
| DeepSeek V4 Flash | 31,650 | 79,050 | 158,150 |

Les estimations sont basées sur les schémas de requêtes moyens observés :
Les estimations sont basées sur les schémas de requêtes observés :

- Grok 4.5 — 1,100 tokens en entrée, 71,500 en cache, 220 tokens en sortie par requête
- GLM-5.2/5.1 — 700 tokens en entrée, 52,000 en cache, 150 tokens en sortie par requête
- Kimi K3/K2.7/K2.6 — 870 tokens en entrée, 55,000 en cache, 200 tokens en sortie par requête
- Kimi K3 — 1,050 tokens en entrée, 76,500 en cache, 300 tokens en sortie par requête
- Kimi K2.7/K2.6 — 870 tokens en entrée, 55,000 en cache, 200 tokens en sortie par requête
- DeepSeek V4 Pro — 750 tokens en entrée, 82,000 en cache, 290 tokens en sortie par requête
- DeepSeek V4 Flash — 790 tokens en entrée, 68,000 en cache, 280 tokens en sortie par requête
- MiniMax M3 — 510 tokens en entrée, 56,000 en cache, 190 tokens en sortie par requête
Expand Down
7 changes: 4 additions & 3 deletions packages/web/src/content/docs/go.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -96,7 +96,7 @@ The table below provides an estimated request count based on typical Go usage pa
| Grok 4.5 | 120 | 300 | 600 |
| GLM-5.2 | 880 | 2,150 | 4,300 |
| GLM-5.1 | 880 | 2,150 | 4,300 |
| Kimi K3 | 140 | 340 | 680 |
| Kimi K3 | 110 | 250 | 490 |
| Kimi K2.7 Code | 1,350 | 4,630 | 9,250 |
| Kimi K2.6 | 1,150 | 2,880 | 5,750 |
| MiMo-V2.5 | 30,100 | 75,200 | 150,400 |
Expand All @@ -109,11 +109,12 @@ The table below provides an estimated request count based on typical Go usage pa
| DeepSeek V4 Pro | 3,450 | 8,550 | 17,150 |
| DeepSeek V4 Flash | 31,650 | 79,050 | 158,150 |

The estimates are based on observed average request patterns:
The estimates are based on observed request patterns:

- Grok 4.5 — 1,100 input, 71,500 cached, 220 output tokens per request
- GLM-5.2/5.1 — 700 input, 52,000 cached, 150 output tokens per request
- Kimi K3/K2.7/K2.6 — 870 input, 55,000 cached, 200 output tokens per request
- Kimi K3 — 1,050 input, 76,500 cached, 300 output tokens per request
- Kimi K2.7/K2.6 — 870 input, 55,000 cached, 200 output tokens per request
- DeepSeek V4 Pro — 750 input, 82,000 cached, 290 output tokens per request
- DeepSeek V4 Flash — 790 input, 68,000 cached, 280 output tokens per request
- MiniMax M3 — 510 input, 56,000 cached, 190 output tokens per request
Expand Down
7 changes: 4 additions & 3 deletions packages/web/src/content/docs/it/go.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -94,7 +94,7 @@ La tabella seguente fornisce una stima del conteggio delle richieste in base a p
| Grok 4.5 | 120 | 300 | 600 |
| GLM-5.2 | 880 | 2,150 | 4,300 |
| GLM-5.1 | 880 | 2,150 | 4,300 |
| Kimi K3 | 140 | 340 | 680 |
| Kimi K3 | 110 | 250 | 490 |
| Kimi K2.7 Code | 1,350 | 4,630 | 9,250 |
| Kimi K2.6 | 1,150 | 2,880 | 5,750 |
| MiMo-V2.5 | 30,100 | 75,200 | 150,400 |
Expand All @@ -107,11 +107,12 @@ La tabella seguente fornisce una stima del conteggio delle richieste in base a p
| DeepSeek V4 Pro | 3,450 | 8,550 | 17,150 |
| DeepSeek V4 Flash | 31,650 | 79,050 | 158,150 |

Le stime si basano sui pattern medi di richieste osservati:
Le stime si basano sui pattern di richieste osservati:

- Grok 4.5 — 1.100 di input, 71.500 in cache, 220 token di output per richiesta
- GLM-5.2/5.1 — 700 di input, 52.000 in cache, 150 token di output per richiesta
- Kimi K3/K2.7/K2.6 — 870 di input, 55.000 in cache, 200 token di output per richiesta
- Kimi K3 — 1.050 di input, 76.500 in cache, 300 token di output per richiesta
- Kimi K2.7/K2.6 — 870 di input, 55.000 in cache, 200 token di output per richiesta
- DeepSeek V4 Pro — 750 di input, 82.000 in cache, 290 token di output per richiesta
- DeepSeek V4 Flash — 790 di input, 68.000 in cache, 280 token di output per richiesta
- MiniMax M3 — 510 di input, 56.000 in cache, 190 token di output per richiesta
Expand Down
7 changes: 4 additions & 3 deletions packages/web/src/content/docs/ja/go.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,7 +86,7 @@ OpenCode Goには以下の制限が含まれています：
| Grok 4.5 | 120 | 300 | 600 |
| GLM-5.2 | 880 | 2,150 | 4,300 |
| GLM-5.1 | 880 | 2,150 | 4,300 |
| Kimi K3 | 140 | 340 | 680 |
| Kimi K3 | 110 | 250 | 490 |
| Kimi K2.7 Code | 1,350 | 4,630 | 9,250 |
| Kimi K2.6 | 1,150 | 2,880 | 5,750 |
| MiMo-V2.5 | 30,100 | 75,200 | 150,400 |
Expand All @@ -99,11 +99,12 @@ OpenCode Goには以下の制限が含まれています：
| DeepSeek V4 Pro | 3,450 | 8,550 | 17,150 |
| DeepSeek V4 Flash | 31,650 | 79,050 | 158,150 |

推定値は、観測された平均的なリクエストパターンに基づいています：
推定値は、観測されたリクエストパターンに基づいています：

- Grok 4.5 — リクエストあたり 入力 1,100トークン、キャッシュ 71,500トークン、出力 220トークン
- GLM-5.2/5.1 — リクエストあたり 入力 700トークン、キャッシュ 52,000トークン、出力 150トークン
- Kimi K3/K2.7/K2.6 — リクエストあたり 入力 870トークン、キャッシュ 55,000トークン、出力 200トークン
- Kimi K3 — リクエストあたり 入力 1,050トークン、キャッシュ 76,500トークン、出力 300トークン
- Kimi K2.7/K2.6 — リクエストあたり 入力 870トークン、キャッシュ 55,000トークン、出力 200トークン
- DeepSeek V4 Pro — リクエストあたり 入力 750トークン、キャッシュ 82,000トークン、出力 290トークン
- DeepSeek V4 Flash — リクエストあたり 入力 790トークン、キャッシュ 68,000トークン、出力 280トークン
- MiniMax M3 — リクエストあたり 入力 510トークン、キャッシュ 56,000トークン、出力 190トークン
Expand Down
7 changes: 4 additions & 3 deletions packages/web/src/content/docs/ko/go.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,7 +86,7 @@ OpenCode Go에는 다음과 같은 한도가 포함됩니다.
| Grok 4.5 | 120 | 300 | 600 |
| GLM-5.2 | 880 | 2,150 | 4,300 |
| GLM-5.1 | 880 | 2,150 | 4,300 |
| Kimi K3 | 140 | 340 | 680 |
| Kimi K3 | 110 | 250 | 490 |
| Kimi K2.7 Code | 1,350 | 4,630 | 9,250 |
| Kimi K2.6 | 1,150 | 2,880 | 5,750 |
| MiMo-V2.5 | 30,100 | 75,200 | 150,400 |
Expand All @@ -99,11 +99,12 @@ OpenCode Go에는 다음과 같은 한도가 포함됩니다.
| DeepSeek V4 Pro | 3,450 | 8,550 | 17,150 |
| DeepSeek V4 Flash | 31,650 | 79,050 | 158,150 |

이 예상치는 관찰된 평균 요청 패턴을 기준으로 합니다.
이 예상치는 관찰된 요청 패턴을 기준으로 합니다.

- Grok 4.5 — 요청당 입력 1,100, 캐시 71,500, 출력 토큰 220
- GLM-5.2/5.1 — 요청당 입력 700, 캐시 52,000, 출력 토큰 150
- Kimi K3/K2.7/K2.6 — 요청당 입력 870, 캐시 55,000, 출력 토큰 200
- Kimi K3 — 요청당 입력 1,050, 캐시 76,500, 출력 토큰 300
- Kimi K2.7/K2.6 — 요청당 입력 870, 캐시 55,000, 출력 토큰 200
- DeepSeek V4 Pro — 요청당 입력 750, 캐시 82,000, 출력 토큰 290
- DeepSeek V4 Flash — 요청당 입력 790, 캐시 68,000, 출력 토큰 280
- MiniMax M3 — 요청당 입력 510, 캐시 56,000, 출력 토큰 190
Expand Down
Loading
Loading