Skip to main content

DeepSeek V4 Flash

DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model from DeepSeek with 284B total parameters and 13B activated parameters, supporting a 1M-token context window. It is designed for fast inference and high-throughput workloads, while maintaining strong reasoning and coding performance. The model includes hybrid attention for efficient long-context processing. Reasoning efforts high and xhigh are supported; xhigh maps to max reasoning. It is well suited for applications such as coding assistants, chat systems, and agent workflows where responsiveness and cost efficiency are important.

Model card on Hugging Face
Input / output price
$0.09 / $0.18 per 1M
Context
1M tokens
Released
Apr 24, 2026
Agent setup

Paste into Claude Code, Cursor, or Codex

Help me set up the Respan gateway: fetch and follow https://www.respan.ai/docs/agent-setup.md. Use model "deepseek/deepseek-v4-flash".

Read docs

Providers

10 providers, cheapest first, retired then OpenRouter last, priced per 1M tokens after gateway discounts.

ProviderInputOutput
DeepInfradeepinfra/deepseek-v4-flash
$0.09/M$0.18/M
Featherlessfeatherless/deepseek-v4-flash
$0.14/M$0.28/M
Parasailparasail/deepseek-v4-flash
$0.14/M$0.28/M
MorphLLMmorphllm/deepseek-v4-flash
$0.141953/M$0.399625/M
Azure Foundryazure_foundry/deepseek-v4-flash
$0.19/M$0.51/M

Is a provider down? DeepSeek status, OpenRouter status

Pricing

Averaged across 10 providers, in USD per 1M tokens after gateway discounts.

Average input
$0.16 per 1M
Average output
$0.38 per 1M
Average cache read
$0.13 per 1M
Average cache write
$0.03 per 1M

Cost by volume

Total cost from 0 to 1T tokens on the cheapest and most expensive provider.

Highest$0.288/MIO.netLowest$0.09/MDeepInfra

Throughput and latency

30-day averages from requests through the Respan gateway. Under each figure, every model on the gateway by its average, this one in blue.

Output speed

Higher is better

102tokens/sacross providers

Faster than 78% of the other 596 models on the gateway

Time to first token

Lower is better

2.0sacross providers

Faster than 43% of the other 153 models on the gateway

Response latency

Lower is better

4.7sacross providers

Faster than 31% of the other 758 models on the gateway

Benchmarks

The Respan Index plus 29 published scores from 10 sources, each linked to where it was published.

45.3/ 100

Respan Index#56 of 176

176 ranked models. The shaded band is this model's 95% interval, 38.9 to 52.3. GLM-5.3-Flash and Claude Opus 4.6 fall inside it, so the gap is within the margin of error.
DeepSeek V4 Flash by category: Coding 29.7, Agents 58.0, Reasoning 46.1, Knowledge & language 35.5, Human preference 55.6
  1. 56DeepSeek V4 Flash45.3
  2. 55GLM-5.3-Flash45.4
  3. 57Claude Opus 4.644.9

Published scores

BenchmarkScore
Terminal Bench 2.1huggingface.co82.7%
Cybergymhuggingface.co76.7%
LiveBench CodingLiveBench75%
WeirdMLHåvard Tveit Ihle63%
DeepSWEhuggingface.co54.4%
NL2Repohuggingface.co54.2%
DeepSWE v1.1Datacurve53.3%
Terminal-Bench 4.0Vals AI18.7%

Compare DeepSeek V4 Flash: DeepSeek V4 Flash vs DeepSeek V4.1 Flash, DeepSeek V4 Flash vs Muse Spark 1.3, DeepSeek V4 Flash vs GPT-6 Luna, DeepSeek V4 Flash vs Claude Sonnet 5.5, DeepSeek V4 Flash vs DeepSeek-V4.1-Flash-1017, DeepSeek V4 Flash vs DeepSeek V4 Pro

Apps

Use DeepSeek V4 Flash in the tools you already run. Each guide points the app at the Respan gateway.

Setup guides for more apps are in the integration docs.

DeepSeek V4 Flash FAQ

Short answers, built from the data on this page.

DeepSeek V4 Flash is a model from Deepseek, released on Apr 24, 2026. DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model from DeepSeek with 284B total parameters and 13B activated parameters, supporting a 1M-token context window. It is designed for fast inference and high-throughput workloads, while maintaining strong reasoning and coding performance. The model includes hybrid attention for efficient long-context processing. Reasoning efforts high and xhigh are supported; xhigh maps to max reasoning. It is well suited for applications such as coding assistants, chat systems, and agent workflows where responsiveness and cost efficiency are important.

On the Respan gateway, DeepSeek V4 Flash costs from $0.09 per 1M input tokens and $0.18 per 1M output tokens on DeepInfra, after gateway discounts. Averaged across its 10 providers, it costs $0.16 per 1M input tokens and $0.38 per 1M output tokens.

DeepSeek V4 Flash is available through 10 providers: DeepInfra, Featherless, Parasail, MorphLLM, Azure Foundry, Tera, and 4 more.

DeepSeek V4 Flash accepts up to 1M tokens of context on its largest provider route.

Point any OpenAI-compatible SDK at https://gateway.respan.ai/v1 with your Respan API key and set the model to "deepseek/deepseek-v4-flash". To set it up with a coding agent, copy the agent setup prompt at the top of this page.

DeepSeek V4 Flash scores 45.3 out of 100 on the Respan Index v1.1, ranking #56 of 176 models. By category: Coding 29.7, Agents 58.0, Reasoning 46.1, Knowledge & language 35.5, and Human preference 55.6. The Respan Index combines independent and lab-reported benchmark results into one score, using benchmark data collected by Respan.

DeepSeek V4 Flash has 29 published benchmark scores, including Terminal Bench 2.1 82.7%, Toolathlon-Verified 70.7%, and GPQA Diamond 91%. Each score links to where it was published; the data is collected by Respan.

Certified where it counts.

Built to meet the security and privacy standards that enterprise and healthcare teams require.

ISO 27001

The internationally recognized standard for information security management.

SOC 2

Secure, compliant management of your data across all of our systems.

GDPR

Operated under GDPR, the world's strictest standard for data privacy.

HIPAA

HIPAA compliant, with a BAA available for healthcare teams.

Works with your entire stack

Use Respan with your favorite frameworks and tools.

AnthropicAI21 LabsxAIVercel AI SDK
LangChainOpenRouterAssemblyAICursor
LiteLLMGroqBasetenMem0
CohereCogneeLlamaIndexFireworks
DeepSeekLinkupHaystackTogether AI
PerplexityInference.netPostHogRubyLLM
SlackMastraAzure OpenAIMistral
AWS BedrockMoonshotHyperspellBAML
ModaInstructorGoogle GeminiParasail
ReductoZapierAgnoNebius
OpenAINovita AIReplicateSuperagent

Built for AI agents.
Break less.
Ship more.