fireworks/deepseek-v4-flash-0731

Common Name: DeepSeek V4 Flash 0731

Released on Jul 31 12:00 AMSupportedTool InvocationSupportedReasoning

DeepSeek's official V4 Flash release for cost-efficient reasoning and agentic workloads.

Context

1000K

Maximum Output

131.1K

Inputtext

Outputtext

Collecting…

Input$0.22/MTokens

Cached Input$0.007/MTokens

Output$0.66/MTokens

Performance Metrics (24h)

$0.30/$1.20/M

ctx512Kmax512Kavail—tps—

InOutCap

MiniMax's reasoning model for long-context coding and agentic tasks.

$0.22/$0.66/M

ctx1.0Mmax384Kavail—tps—

InOutCap

DeepSeek's cost-efficient hybrid-thinking model in the V4 family.

$0.40/$1.60/M

ctx262Kmax66Kavail—tps—

InOutCap

Alibaba's multimodal reasoning model for general-purpose and agentic workloads.

$0.30/$1.20/M

ctx197Kmax197Kavail—tps—

InOutCap

MiniMax's reasoning model for coding and agentic tasks.