LLM Reference

DeepSeek V4 Flash

Released
2026-04-24
Last refreshed
2026-07-03
Status
Researched 19d ago
Open sourceCommercial use: permittedCodingRAGAgentsLong contextClassificationJSON / Tool use

DeepSeek V4 Flash is worth evaluating for coding, rag, and agents when its provider route and context window match the workload.

Use it for

  • Teams evaluating coding, rag, and agents
  • Workloads that can use a 1m context window
  • Buyers comparing 4 tracked provider routes

Do not use it for

  • Vision or document-understanding workloads
Specifications
Released
2026-04-24
Context
1m
Max output
384,000
Parameters
284B
Architecture
Mixture of Experts
Specialization
general
Openness
Open source
License
MITOSI-approvedCommercial use: permitted
Weights
Available
Code
Unknown
Training
Pretrained
Created by

Advancing artificial general intelligence (AGI).

Hangzhou, Zhejiang, China
Founded 2023
Website
Pricing
Output / 1M
$0.180
Input / 1M
$0.090

Cheapest of 5 routes · OpenRouter · cache read $0.018

About

DeepSeek V4 Flash is a 284B parameter (13B activated) Mixture-of-Experts language model with 1M-token context. Features a hybrid attention architecture combining Compressed Sparse Attention (CSA) and Heavily Compressed Attention (HCA) for efficient long-context inference. Supports thinking and non-thinking modes. Legacy API aliases deepseek-chat and deepseek-reasoner map to this model's non-thinking and thinking modes respectively. Pricing: $0.14/1M input, $0.28/1M output (cache hit: $0.0028/1M input). MIT licensed.

DeepSeek V4 Flash is an open-source model in the DeepSeek V4 family. The structured metadata tracks a 1m-token context window, reasoning, function calling, tool use, and structured outputs. This page tracks provider routes through DeepSeek Platform, OpenRouter, Microsoft Foundry, and 2 more, with the cheapest tracked route listed at $0.09 input and $0.18 output per 1M tokens. Headline tracked benchmarks include Google-Proof Q&A 88.1, MMLU PRO 86.4, and SWE-bench Verified 79.0.

Top use-case fit: coding, agents, and build tasks

Coding

Q/$ B

4 relevant benchmarks in the decision map.

RAG

Included by capability and metadata signals in the decision map.

Agents

Q/$ A

1 relevant benchmark in the decision map.

Provider price ladder

Compare all 5

Compare API pricing across 4 providers for input and output tokens, batch, and cached reads when available.

ProviderInput / 1MOutput / 1MCacheRoute
OpenRouter$0.090$0.180read $0.018
Serverless
DeepSeek Platform$0.140$0.280read $0.0028
Serverless
Novita AI$0.140$0.280read $0.028
Serverless
Vercel AI Gateway$0.140$0.280read $0.0028
Serverless

Available via routers & gateways(8)

Capabilities

ReasoningFunction CallingTool UseStructured OutputsPrompt Caching

Benchmark peer barsfor Coding

Benchmark scores(16)

Scores are benchmark-specific and are direction-aware: the same numeric gap can mean very different outcomes across suites. Use the leaderboard context and this model's provider route to decide whether the winning margin is meaningful for your workload.
BenchmarkScoreVersionEvaluationSource
Google-Proof Q&A88.1GPQA Diamond pass@1, V4-Flash Think MaxObserved 2026-07-03Source
MMLU PRO86.4MMLU-Pro EM, V4-Flash Think HighObserved 2026-07-03Source
SWE-bench Verified79.0SWE-bench Verified resolved, V4-Flash Think MaxObserved 2026-07-03Source
SWE-bench Pro52.6SWE-bench Pro resolved, V4-Flash Think MaxObserved 2026-07-03Source
LiveCodeBench91.6LiveCodeBench pass@1, V4-Flash Think MaxObserved 2026-07-03Source
HumanEval69.5HumanEval pass@1, DeepSeek-V4-Flash-Base, 0-shotObserved 2026-07-03Source
Massive Multitask Language Understanding88.7MMLU EM, DeepSeek-V4-Flash-Base, 5-shotObserved 2026-07-03Source
Terminal-Bench 2.056.9Terminal Bench 2.0 accuracy, V4-Flash Think MaxObserved 2026-07-03Source
GeneBench-Pro2.4xhighObserved 2026-06-30Source
Humanity's Last Exam34.8Humanity's Last Exam pass@1, no tools, V4-Flash Think MaxObserved 2026-07-03Source
Humanity's Last Exam — With Tools45.1Humanity's Last Exam with tools pass@1, V4-Flash Think MaxObserved 2026-07-03Source
Chatbot Arena1437.0LMArena Chatbot Arena text leaderboard Elo, deepseek-v4-flash-thinkingObserved 2026-07-03Source
SWE-bench Multilingual73.3SWE-bench Multilingual resolved, V4-Flash Think MaxObserved 2026-07-03Source
MCP-Atlas69.0MCPAtlas pass@1, V4-Flash Think MaxObserved 2026-07-03Source
BrowseComp73.2BrowseComp pass@1, V4-Flash Think MaxObserved 2026-07-03Source
Mathematics Aptitude Test of Heuristics57.4MATH EM, DeepSeek-V4-Flash-Base, 4-shotObserved 2026-07-03Source

Migration checks

No linked migration route is available for this model yet.

Compare DeepSeek V4 Flash with other models

Show all 69 popular comparisonssorted by 7-day search impressions
DeepSeek V4 Flash vs Kimi K2.51KDeepSeek V4 Flash vs Claude Haiku 4.51KDeepSeek V4 Flash vs GLM-5725DeepSeek V4 Flash vs GPT-5.2 Codex634DeepSeek V4 Flash vs DeepSeek R1602DeepSeek V4 Flash vs Qwen3.5-27B571DeepSeek V4 Flash vs Qwen3.5-397B-A17B552DeepSeek V4 Flash vs GPT-5.4519DeepSeek V4 Flash vs Qwen3.5-122B-A10B511DeepSeek V4 Flash vs Kimi K2 Instruct439DeepSeek V4 Flash vs Claude Opus 4.6429DeepSeek V4 Flash vs DeepSeek V3.2410DeepSeek V4 Flash vs Llama 3 70B Instruct407DeepSeek V4 Flash vs GPT-5.5340DeepSeek V4 Flash vs Llama 3.2 1B Instruct329DeepSeek V4 Flash vs GLM-5V-Turbo317DeepSeek V4 Flash vs Claude 3.7 Sonnet313DeepSeek V4 Flash vs Qwen3.5-9B292DeepSeek V4 Flash vs Qwen2.5-72B-Instruct289DeepSeek V4 Flash vs DeepSeek R1 0528266DeepSeek V4 Flash vs o3252DeepSeek V4 Flash vs GPT-4o-mini Search Preview247DeepSeek V4 Flash vs Qwen3.5-35B-A3B240DeepSeek V4 Flash vs Mistral Large 2206DeepSeek V4 Flash vs Qwen2.5-72B200DeepSeek V4 Flash vs Claude Opus 4.5168DeepSeek V4 Flash vs Claude Mythos Preview166DeepSeek V4 Flash vs Trinity-Large-Thinking160DeepSeek V4 Flash vs Llama 3 8B Instruct157DeepSeek V4 Flash vs Grok 3 Mini144DeepSeek V4 Flash vs Grok Build 0.1138DeepSeek V4 Flash vs Llama 2 13B Chat113DeepSeek V4 Flash vs Qwen3-235B-A22B112DeepSeek V4 Flash vs Mistral Nemotron108DeepSeek V4 Flash vs GPT-5.2105DeepSeek V4 Flash vs DeepSeek V3 Base102DeepSeek V4 Flash vs Kimi K2 Thinking Turbo96DeepSeek V4 Flash vs Mistral Large 3 675B Instruct85DeepSeek V4 Flash vs o3 Mini76DeepSeek V4 Flash vs Gemma 2 2B65DeepSeek V4 Flash vs Grok-364DeepSeek V4 Flash vs Gemini 2.5 Flash Live API61DeepSeek V4 Flash vs Llama 3.2 1B56DeepSeek V4 Flash vs Gemma 7B Instruct55DeepSeek V4 Flash vs Together AI Qwen2-7B-Instruct52DeepSeek V4 Flash vs DeepSeek R1 Basic49DeepSeek V4 Flash vs Trinity-Large-Preview45DeepSeek V4 Flash vs Llama 3.1 70B Instruct42DeepSeek V4 Flash vs Qwen2-7B-Instruct35DeepSeek V4 Flash vs Phi-3 Mini 4k34DeepSeek V4 Flash vs GPT-5.5 Instant30DeepSeek V4 Flash vs GPT-5.4-Cyber28DeepSeek V4 Flash vs Qwen3.6 Max Preview27DeepSeek V4 Flash vs Llama 3.1 405B Instruct27DeepSeek V4 Flash vs Gemini 2.5 Pro Computer Use Preview21DeepSeek V4 Flash vs Mixtral 8x7B19DeepSeek V4 Flash vs Gemma 2 9B SahabatAI Instruct18DeepSeek V4 Flash vs Phi-4 Reasoning Vision 15B16DeepSeek V4 Flash vs Together AI - Llama 3 8B Lite13DeepSeek V4 Flash vs Magistral Small 250610DeepSeek V4 Flash vs Gemini 3.1 Flash-Lite10DeepSeek V4 Flash vs Code Davinci 0018DeepSeek V4 Flash vs ShieldGemma 9B8DeepSeek V4 Flash vs Phi-4 Mini Flash Reasoning8DeepSeek V4 Flash vs DeepSeek R1 Lite8DeepSeek V4 Flash vs Mixtral 8x22B Instruct v0.37DeepSeek V4 Flash vs GPT-5.5 Pro7DeepSeek V4 Flash vs o3 Deep Research6DeepSeek V4 Flash vs DeepSeek V30

Frequently asked questions

What is the context window of DeepSeek V4 Flash?

DeepSeek V4 Flash has a context window of 1m tokens.

What is the max output of DeepSeek V4 Flash?

DeepSeek V4 Flash can generate up to 384,000 output tokens.

How much does DeepSeek V4 Flash cost?

DeepSeek V4 Flash pricing ranges from $0.09/1M to $0.19/1M input tokens depending on the provider.

When was DeepSeek V4 Flash released?

DeepSeek V4 Flash was released on 2026-04-24.

Which providers offer DeepSeek V4 Flash?

DeepSeek V4 Flash is available from 5 providers: DeepSeek Platform, OpenRouter, Microsoft Foundry, Vercel AI Gateway, Novita AI.

What benchmarks has DeepSeek V4 Flash been tested on?

DeepSeek V4 Flash has been evaluated on 16 benchmarks, including Google-Proof Q&A, MMLU PRO, SWE-bench Verified, SWE-bench Pro, LiveCodeBench.