DeepSeek V4 Pro and Flash land on Vercel AI Gateway
Vercel AI Gateway now supports two DeepSeek V4 model variants — V4 Pro and V4 Flash — both with a 1M token context window, accessible via the AI SDK.
Score breakdown
Developers building agentic coding pipelines or MCP-based workflows can now route DeepSeek V4 Pro or Flash through Vercel AI Gateway's unified API, gaining built-in observability, failover, and cost tracking without additional infrastructure.
- 01DeepSeek V4 is now available on Vercel AI Gateway in two variants: V4 Pro and V4 Flash.
- 02Both models have a 1M token context window by default.
- 03V4 Pro targets agentic coding, formal mathematical reasoning, and long-horizon workflows.
Vercel AI Gateway has added support for DeepSeek V4, available in two variants: DeepSeek V4 Pro and DeepSeek V4 Flash. Both models default to a 1M token context window. V4 Pro is designed for agentic coding tasks — including feature development, bug fixing, and refactoring across stacks — as well as formal mathematical reasoning and long-horizon workflows. It supports tool use across harnesses such as MCP workflows and agent frameworks, and is also capable of producing well-structured long-form documents.
The release notes describe it as a strong fit for high-volume and latency-sensitive use cases.
V4 Flash is positioned as a lighter alternative that performs close to V4 Pro on reasoning tasks and holds up on simpler agent workloads, but with a smaller parameter size that enables faster responses and lower API cost. The release notes describe it as a strong fit for high-volume and latency-sensitive use cases.
Both models are accessible via the AI SDK by setting the model to `deepseek/deepseek-v4-pro` or `deepseek/deepseek-v4-flash`. Vercel AI Gateway provides a unified API layer for calling models, tracking usage and cost, and configuring retries, failover, and performance optimizations. It also includes custom reporting, observability, Bring Your Own Key support, and intelligent provider routing with automatic retries.
Key facts
- 01DeepSeek V4 is now available on Vercel AI Gateway in two variants: V4 Pro and V4 Flash.
- 02Both models have a 1M token context window by default.
- 03V4 Pro targets agentic coding, formal mathematical reasoning, and long-horizon workflows.
- 04V4 Pro supports tool use across MCP workflows and agent frameworks.
- 05V4 Flash offers near-V4 Pro reasoning performance with a smaller parameter size for faster, cheaper responses.
- 06Models are accessed via the AI SDK using `deepseek/deepseek-v4-pro` or `deepseek/deepseek-v4-flash`.
- 07AI Gateway includes custom reporting, observability, Bring Your Own Key support, and intelligent provider routing with automatic retries.