WWW Agents
Create Agent
Try DeepSeek V4 onlineVerified August 1, 2026

Chat with DeepSeek V4 Flash or Pro

Use V4 Flash for fast, economical conversations, or choose V4-Pro for demanding coding, reasoning, and tool-connected work. Either model can power a private Agent.

1M contextThinking and non-thinking modesTool calls
Chat with DeepSeek V4 FlashChat with DeepSeek V4-Pro

Choose DeepSeek V4 Flash or Pro

Chat with either shared model directly, or create a new private Agent. Flash is the faster, lower-cost choice; Pro is for harder coding and reasoning tasks.

Fixed model: DeepSeek V4 Flash

Chat now or create an Agent

Start a conversation with the fixed model, or use it to create a private Agent with your own Skill and instructions.

Chat with DeepSeek V4 FlashCreate Agent with DeepSeek V4 Flash
Fixed model: DeepSeek V4-Pro

Chat now or create an Agent

Start a conversation with the fixed model, or use it to create a private Agent with your own Skill and instructions.

Chat with DeepSeek V4-ProCreate Agent with DeepSeek V4-Pro

Release and API status

Current public facts from DeepSeek documentation.

API model IDs
DeepSeek documents deepseek-v4-pro and deepseek-v4-flash on the same base URLs for OpenAI and Anthropic formats.
Legacy model retirement
deepseek-chat and deepseek-reasoner were retired on July 24, 2026 at 15:59 UTC and are no longer valid migration targets.
Verification date
Facts and prices were checked against DeepSeek docs on August 1, 2026.

DeepSeek V4 Flash and Pro facts

Compare the shared V4 capabilities, then choose Flash for speed and cost or Pro for harder Agent workflows.

Model ID
deepseek-v4-pro

Use deepseek-v4-pro for the Pro variant; deepseek-v4-flash is the lower-cost companion model.

Context window
1M

DeepSeek documents 1M context for both V4-Pro and V4-Flash.

Max output
384K

DeepSeek lists a maximum output of 384K tokens.

Modalities
Text

The pricing table documents text chat capabilities, tool calls, JSON output, and FIM beta.

Reasoning
Thinking and non-thinking

Both V4 models support thinking and non-thinking modes.

Features
JSON, tools, prefix, FIM

DeepSeek lists JSON output, tool calls, chat prefix completion, and FIM completion beta.

Best-fit Agent use cases

Prioritize workloads that benefit from long context and low token cost.

Agentic coding
Use V4-Pro for repository analysis, code generation, and tool-connected coding workflows.
Long-document reasoning
Use the 1M context window for large specifications, logs, and source packs.
High-volume private Agents
Use Pro for harder tasks and compare Flash when throughput cost matters.

Decision evidence

Provider-published facts that matter for Agent selection.

DimensionOfficial evidenceDecision use
Agentic capabilitiesDeepSeek describes V4-Pro as optimized for agentic coding and integrated with leading coding agents.Shortlist it for coding and tool-loop workflows, then test on your own acceptance scenarios.
Long contextDeepSeek documents 1M context and 384K max output.Useful when the Agent must hold large repositories or documents in one request.
Token priceDeepSeek publishes low cache-hit, cache-miss, and output prices for V4-Pro and V4-Flash.Measure successful-task cost, especially when repeated long prompts benefit from cache hits.

DeepSeek release claims are provider-published; validate with your own tasks before production use.

Published API pricing

Rates per 1M tokens from DeepSeek pricing.

ModelInputOutput
deepseek-v4-pro$0.003625 hit / $0.435 miss$0.87
deepseek-v4-flash$0.0028 hit / $0.14 miss$0.28

Prices are per 1M tokens. Product prices may vary, and DeepSeek recommends checking the pricing page regularly.

Operational limits

Validate these before switching a production Agent.

Legacy IDs are retired
deepseek-chat and deepseek-reasoner are retired. Use deepseek-v4-flash or deepseek-v4-pro for new integrations.
Text-first API profile
The documented pricing table centers on chat, tools, JSON output, prefix completion, and FIM beta.
Price can change
DeepSeek explicitly notes product prices may vary, so store the verification date.

Official sources

Primary DeepSeek documentation used for this page.

DeepSeek changelogV4 model IDs, API availability, and legacy retirement notice.DeepSeek V4 preview releaseProvider-published positioning, long-context, and agent capability notes.DeepSeek models and pricingModel details, context, output, features, and token rates.

DeepSeek V4 Flash and Pro FAQ

Answers about DeepSeek V4 online Chat, Flash versus Pro, API pricing, context, and Agent creation.

Start with the right DeepSeek V4 model

Chat with Flash for speed and lower cost, use Pro for harder work, or create a private DeepSeek V4 Agent.

Chat with DeepSeek V4-ProCreate Agent with DeepSeek V4-Pro