deepseek logo
deepseek

DeepSeek: DeepSeek V4 Pro 0813

Description de la source (anglais)

DeepSeek V4 Pro 0813 is a large-scale mixture-of-experts model from DeepSeek. This is the GA release of DeepSeek V4 Pro.

Présentation

Caractéristiques du modèle

Contexte
1 048 576 tokens
Sortie maximale
384 000 tokens
Architecture
text->text
Tokenizer
DeepSeek
Limite des connaissances
Non indiqué
Modéré
Non
OPENROUTER

Tarification complète

Tarifs synchronisés depuis OpenRouter, par million de tokens.

Entrée
$0.4987
par million de tokens
Sortie
$1.4961
par million de tokens
Lecture du cache
$0.0166
par million de tokens
API

Configuration du modèle

Paramètres par défaut

temperature
1
top_p
1

Raisonnement

Raisonnement obligatoire
Non
Paramètres par défaut
high
Capacités et modalités
max, high, low
API

Démarrage rapide

Définissez OPENROUTER_API_KEY localement. Python nécessite requests ; JavaScript s’exécute dans Node.js. Gardez la clé côté serveur.

Documentation de l’API

curl --fail-with-body https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek/deepseek-v4-pro-0813","messages":[{"role":"user","content":"Hello!"}]}'
Capacités et modalités

EntréeSortie

Entrée
text
Sortie
text
Raisonnement

Non · max · high · low

API

Paramètres API pris en charge

frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p
Fournisseurs disponibles

24 fournisseurs

Vérifié: 23 septembre 2026

Fournisseurs en direct sur OpenRouter

Disponibilité, latence, débit et routage évoluent en continu. Consultez la source pour les données en temps réel.

OpenRouter

Baidu

fp8
Contexte
1.0M
Sortie maximale
393K
Entrée
$0.4
Sortie
$1.1999
Lecture du cache
$0.0127
Écriture du cache
Non indiqué

Sail Research

fp4
Contexte
1.0M
Sortie maximale
384K
Entrée
$0.4
Sortie
$4.3
Lecture du cache
$0.033
Écriture du cache
Non indiqué

Sail Research

fp4
Contexte
1.0M
Sortie maximale
384K
Entrée
$0.4
Sortie
$4.3
Lecture du cache
$0.033
Écriture du cache
Non indiqué

StreamLake

Non indiqué
Contexte
1.0M
Sortie maximale
384K
Entrée
$0.462
Sortie
$1.386
Lecture du cache
$0.0154
Écriture du cache
Non indiqué

Ionstream

Non indiqué
Contexte
1.0M
Sortie maximale
944K
Entrée
$0.4992
Sortie
$2.304
Lecture du cache
$0.0352
Écriture du cache
Non indiqué

Wafer

Non indiqué
Contexte
1.0M
Sortie maximale
944K
Entrée
$0.5
Sortie
$2.9
Lecture du cache
$0.25
Écriture du cache
Non indiqué

DeepSeek

Non indiqué
Contexte
1.0M
Sortie maximale
384K
Entrée
$0.66
Sortie
$1.98
Lecture du cache
$0.022
Écriture du cache
Non indiqué

Novita

fp8
Contexte
1.0M
Sortie maximale
393K
Entrée
$0.99
Sortie
$2.97
Lecture du cache
$0.033
Écriture du cache
Non indiqué

NextBit

fp8
Contexte
1.0M
Sortie maximale
944K
Entrée
$1.056
Sortie
$3.168
Lecture du cache
$0.035
Écriture du cache
Non indiqué

GMICloud

fp8
Contexte
1.0M
Sortie maximale
944K
Entrée
$1.056
Sortie
$3.168
Lecture du cache
$0.0352
Écriture du cache
Non indiqué

Alibaba

Non indiqué
Contexte
1M
Sortie maximale
393K
Entrée
$1.122
Sortie
$3.366
Lecture du cache
$0.1122
Écriture du cache
Non indiqué

DeepInfra

fp8
Contexte
1.0M
Sortie maximale
16K
Entrée
$1.3
Sortie
$2.6
Lecture du cache
$0.1
Écriture du cache
Non indiqué

CoreWeave

fp8
Contexte
1.0M
Sortie maximale
944K
Entrée
$1.31
Sortie
$3.96
Lecture du cache
$0.044
Écriture du cache
Non indiqué

AtlasCloud

fp8
Contexte
1.0M
Sortie maximale
393K
Entrée
$1.32
Sortie
$3.96
Lecture du cache
$0.132
Écriture du cache
Non indiqué

BaseTen

fp4
Contexte
1.0M
Sortie maximale
262K
Entrée
$1.32
Sortie
$3.96
Lecture du cache
$0.132
Écriture du cache
Non indiqué

Parasail

fp8
Contexte
1.0M
Sortie maximale
944K
Entrée
$1.32
Sortie
$3.96
Lecture du cache
$0.044
Écriture du cache
Non indiqué

Together

Non indiqué
Contexte
1.0M
Sortie maximale
944K
Entrée
$1.32
Sortie
$3.96
Lecture du cache
$0.13
Écriture du cache
Non indiqué

DigitalOcean

Non indiqué
Contexte
1.0M
Sortie maximale
944K
Entrée
$1.32
Sortie
$3.96
Lecture du cache
$0.044
Écriture du cache
Non indiqué

SiliconFlow

fp8
Contexte
1.0M
Sortie maximale
393K
Entrée
$1.32
Sortie
$3.96
Lecture du cache
$0.044
Écriture du cache
Non indiqué

BaseTen

fp4
Contexte
1.0M
Sortie maximale
262K
Entrée
$1.32
Sortie
$3.96
Lecture du cache
$0.132
Écriture du cache
Non indiqué

Cloudflare

Non indiqué
Contexte
1.0M
Sortie maximale
944K
Entrée
$1.32
Sortie
$3.96
Lecture du cache
$0.044
Écriture du cache
Non indiqué

Fireworks

Non indiqué
Contexte
1.0M
Sortie maximale
944K
Entrée
$1.32
Sortie
$3.96
Lecture du cache
$0.044
Écriture du cache
Non indiqué

Phala

Non indiqué
Contexte
1.0M
Sortie maximale
393K
Entrée
$1.45
Sortie
$4.36
Lecture du cache
$0.15
Écriture du cache
Non indiqué

Venice

Non indiqué
Contexte
1M
Sortie maximale
33K
Entrée
$1.65
Sortie
$4.95
Lecture du cache
$0.165
Écriture du cache
Non indiqué
deepseek

4 modèles

Tous les modèles
deepseek logodeepseek

DeepSeek: DeepSeek V4.1 Flash

DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...

Contexte
1.0M
Entrée
text · image
Sortie
text
Entrée: $0.15Sortie: $0.6par million de tokens
Voir la fiche du modèle
deepseek logodeepseek

DeepSeek: DeepSeek V4.1 Flash (batch)

DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...

Contexte
1.0M
Entrée
text · image
Sortie
text
Entrée: $0.112Sortie: $0.336par million de tokens
Voir la fiche du modèle
deepseek logodeepseek

DeepSeek: DeepSeek V4 Flash Vision Exp

DeepSeek V4 Flash Vision Exp is an experimental vision-enabled version of DeepSeek V4 Flash 0731 from DeepSeek, adding image understanding while matching the base model on text capabilities including agents,...

Contexte
1.0M
Entrée
text · image
Sortie
text
Entrée: $0.22Sortie: $0.66par million de tokens
Voir la fiche du modèle
deepseek logodeepseek

DeepSeek: DeepSeek V4 Flash 0731

DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows....

Contexte
1.3M
Entrée
text
Sortie
text
Entrée: $0.04Sortie: $0.64par million de tokens
Voir la fiche du modèle