II.
ModelVersion overview
Reference · livemodel:deepseek-v4-flash@current
DeepSeek V4 Flash overview
Inspect the raw attributes, linked wiki pages, and inbound or outbound graph edges for model:deepseek-v4-flash@current.
Attributes
displayName
DeepSeek V4 Flash
costPerMTokInput
0.14
costPerMTokOutput
0.28
supportsExtendedThinking
true
supportsThinkingBudgetTokens
false
modelFamilyId
versionRange
>=4.0.0 <5.0.0
kind
reasoning
contextWindowTokens
1000000
maxOutputTokens
384000
modalities
- text
modalityDirections
- kindtextdirectioninput
- kindtextdirectionoutput
releaseDate
2026-04-24
lifecycleStatus
preview
lifecycleStatusChangedAt
2026-04-24
modelCardUrl
supportsFineTuning
false
providerModelIds
deepseekApi
deepseek-v4-flash
deepseekVersion
DeepSeek-V4-Flash
regions
- global
Outgoing edges
belongs_to_family1
- model-family:deepseek·ModelFamilyDeepSeek
realizes1
- layer:1-model·LayerModel
served_by3
- provider:deepseek·ProviderDeepSeek
- provider:together-ai·ModelProviderProductTogether AI
- provider:fireworks-ai·ModelProviderProductFireworks AI
speaks2
- model-transport:openai-compat·ModelTransportProtocolOpenAI-compatible chat completions
- model-transport:anthropic-messages·ModelTransportProtocolAnthropic Messages
supersedes1
- model:deepseek-v3@current·ModelVersionDeepSeek V3
supports7
- capability:streaming·CapabilityStreaming
- capability:supports-tool-use·CapabilitySupports tool use
- capability:supports-json-mode·CapabilitySupports JSON mode
- capability:supports-prompt-caching·CapabilitySupports prompt caching
- capability:supports-thinking·CapabilitySupports thinking
- capability:extended-thinking·CapabilityExtended thinking
- capability:can-multi-turn·CapabilityCan multi-turn
version_of1
- model-family:deepseek·ModelFamilyDeepSeek
Incoming edges
about_subject21
- claim:model-version-model-deepseek-v4-flash-current-model-code·Claim
- claim:model-version-model-deepseek-v4-flash-current-version-name·Claim
- claim:model-version-model-deepseek-v4-flash-current-versionrange·Claim
- claim:model-version-model-deepseek-v4-flash-current-kind·Claim
- claim:model-version-model-deepseek-v4-flash-current-contextwindowtokens·Claim
- claim:model-version-model-deepseek-v4-flash-current-maxoutputtokens·Claim
- claim:model-version-model-deepseek-v4-flash-current-costpermtokinput·Claim
- claim:model-version-model-deepseek-v4-flash-current-costpermtokcachedinput·Claim
- claim:model-version-model-deepseek-v4-flash-current-costpermtokoutput·Claim
- claim:model-version-model-deepseek-v4-flash-current-thinking-modes·Claim
- claim:model-version-model-deepseek-v4-flash-current-supportsextendedthinking·Claim
- claim:model-version-model-deepseek-v4-flash-current-supportsthinkingbudgettokens·Claim
- claim:model-version-model-deepseek-v4-flash-current-modalities·Claim
- claim:model-version-model-deepseek-v4-flash-current-modalitydirections·Claim
- claim:model-version-model-deepseek-v4-flash-current-lifecyclestatus·Claim
- claim:model-version-model-deepseek-v4-flash-current-releasedate·Claim
- claim:model-version-model-deepseek-v4-flash-current-modelcardurl·Claim
- claim:model-version-model-deepseek-v4-flash-current-supportsfinetuning·Claim
- claim:model-version-model-deepseek-v4-flash-current-regions·Claim
- claim:model-version-model-deepseek-v4-flash-current-speaks·Claim
- claim:model-version-model-deepseek-v4-flash-current-json-tool-prefix-fim·Claim
has_version1
- model-family:deepseek·ModelFamilyDeepSeek
serves3
- provider:deepseek·ProviderDeepSeek
- provider:fireworks-ai·ModelProviderProductFireworks AI
- provider:together-ai·ModelProviderProductTogether AI