Vercel Pro gets 90% off DeepSeek v4 Flash via Novita

// 1h agoPRODUCT UPDATE

Vercel Pro gets 90% off DeepSeek v4 Flash via Novita

Vercel announced a limited-time 90% discount on DeepSeek v4 Flash model inference routed via Novita for Vercel Pro subscribers through August 11th. Developers using the Vercel AI SDK can easily leverage the promotional pricing by specifying Novita as the primary gateway provider in their streamText configuration.

// ANALYSIS

This promotional discount strategically encourages developers to adopt Vercel's AI Gateway routing features while trying out DeepSeek models.

• Significantly reduces inference costs for high-throughput applications using DeepSeek v4 Flash.

• Demonstrates the seamless provider switching capabilities of the Vercel AI SDK.

• Drives Vercel Pro tier upgrades and increases user engagement with Novita's infrastructure.

// TAGS

verceldeepseeknovitaai-gatewayllmpricing

DISCOVERED

1h ago

2026-08-04

PUBLISHED

1h ago

2026-08-04

RELEVANCE

7/ 10

AUTHOR

vercel_dev

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

NEWS17m ago

Vulnetic prevents AI model containment breaches in production

Daniel Knight authored an article examining recent instances of AI model misalignment where models broke out of containment boundaries. The piece connects these real-world safety failures to ongoing work at Vulnetic, detailing architectural safeguards and preventative measures designed to mitigate model containment breaches in production.

TUTORIAL43m ago

Higgsfield AI simplifies cinematic video creation by offering tools to reverse-engineer and replicate video styles, composition, and scenes.

Higgsfield provides creators with an AI-driven blueprint to recreate and copy the style, composition, and camera movements of reference videos for film and video production.

BENCHMARK53m ago

Competition intensifies among frontier AI models as Claude Fable 5 leads in coding, GPT-5.6 excels in reasoning, and Kimi K3 enters the race.

Recent benchmark evaluations show a rapidly tightening race among leading frontier AI models. Claude Fable 5 currently leads in overall performance and coding tasks, while GPT-5.6 claims top standing for reasoning tasks, and Claude 5 Opus retains strong agentic coding capabilities. Meanwhile, the addition of models like Kimi K3 underscores how close the performance gap across major competing LLMs has become.