https://vercel.com/blog
DeepSignal tracks AI updates from Vercel AI, filtering research and product signals into plain-English summaries, signal scores and source-linked article pages.
Current topics: Tooling, Open Source, AI Coding, Agent, AI Assistant · Companies: Vercel, Claude, DeepSeek, Gemini
High-signal updates

Google's Nano Banana 2 Lite is now available on AI Gateway, offering fast, multimodal image and text generation under 4 seconds at a cost of $0.034 per 1K images, significantly cheaper than its predecessors. This model supports advanced features like editing images over multiple turns and integrates seamlessly with the AI SDK.
The launch of Google's Nano Banana 2 Lite on AI Gateway, offering fast multimodal image and text generation at $0.034 per 1K images, significantly reduces costs for developers and product managers. This enables more innovative applications and features, enhancing user experience while optimizing budget allocations for AI projects.

Claude Sonnet 5 from Anthropic is now available on Vercel AI Gateway, enhancing coding and agentic tasks with improved instruction adherence and document parsing. Launch pricing is set at $2 per million input tokens and $10 per million output tokens until August 31, 2026, with a standard price of $3/M and $15/M thereafter.
The availability of Claude Sonnet 5 on Vercel AI Gateway enhances coding and agentic tasks with better instruction adherence and document parsing, which can improve developer productivity and application performance. Builders and PMs should consider integrating this model for more efficient workflows, while investors may see this as a signal of growing demand for advanced AI capabilities in software development.

Vercel AI's Gateway now supports real-time voice capabilities, enabling two-way voice agents, text-to-speech, and speech-to-text functionalities using models from OpenAI and xAI. This feature, currently in beta with AI SDK 7, allows seamless integration of audio alongside existing text, image, and video modalities.
Vercel AI's Gateway now supports real-time voice capabilities, allowing builders and PMs to create more interactive applications that integrate voice with existing text, image, and video functionalities. This development opens up new avenues for user engagement and can enhance product offerings, making it a significant consideration for investors looking at innovative tech solutions.

xAI's audio models, including real-time voice (xai/grok-voice-think-fast-1.0), text-to-speech (xai/grok-tts), and speech-to-text (xai/grok-stt), are now available on Vercel AI Gateway. These models can be integrated via the AI SDK 7, providing robust routing and observability for developers.
The availability of xAI's audio models on Vercel AI Gateway allows builders to easily integrate advanced voice capabilities, such as real-time voice and text-to-speech, into their applications. This development streamlines the implementation process and enhances user experience, making it a significant opportunity for PMs and investors looking to capitalize on AI-driven audio solutions.
AI Gateway by Vercel now supports real-time voice, speech, and transcription capabilities, allowing developers to create voice agents and transcribe audio seamlessly. The features include real-time voice agents, text-to-speech, and speech-to-text functionalities, all available in beta through AI SDK 7 without additional fees.
Vercel's AI Gateway now supports real-time voice, speech, and transcription capabilities, enabling developers to build advanced voice agents and transcription services without extra costs. This development allows builders and PMs to enhance user engagement and accessibility in their applications, while investors can recognize the potential for new revenue streams in the voice technology market.

Vercel's AI SDK 7, with over 16 million weekly downloads, enhances agent development by standardizing model reasoning, introducing typed tool contexts, and optimizing file handling for improved performance. It supports durable workflows, tool approvals, and offers a terminal UI for rapid testing.
Vercel's AI SDK 7 introduces standardized model reasoning and typed tool contexts, which streamline agent development and improve performance. This is crucial for builders and PMs looking to enhance the efficiency of their AI applications, while investors should note the SDK's popularity, as indicated by its 16 million weekly downloads, signaling strong market demand.

Vercel's innovative approach to teaching coding agents product design involves integrating accepted product decisions into their code repository, allowing agents to learn from context and maintain design consistency. This system, which includes structured routing and deterministic lint rules, enhances the agents' ability to produce UI that aligns with established design standards while ensuring traceability and human oversight in the review process.
Vercel's integration of product design principles into coding agents represents a significant advancement in automating UI development, ensuring that generated designs adhere to established standards while maintaining traceability. This development is crucial for builders and PMs as it streamlines the design process, reduces errors, and enhances collaboration between human and AI, making it an attractive proposition for investors focused on innovative tech solutions.

Vercel AI has introduced Deep Agents and OpenCode adapters in the AI SDK Harness, enabling seamless switching between coding-agent runtimes without altering application code. These additions enhance the LangChain runtime with built-in tools and multi-turn session capabilities, supporting various model configurations.
The introduction of Deep Agents and OpenCode in Vercel AI's SDK Harness allows developers to easily switch coding-agent runtimes without modifying their application code, streamlining development processes. This flexibility can significantly reduce time-to-market for AI-driven applications, making it an attractive proposition for builders, PMs, and investors looking to leverage advanced AI capabilities.
AI SDK 7 from Vercel introduces a comprehensive platform for building production agents in TypeScript, supporting multiple modalities like text, audio, and video. Key features include a new harness layer for integrating existing coding agents and enhanced UI for reliable streaming and message handling.
The release of AI SDK 7 by Vercel is significant for builders and PMs as it enables the development of versatile production agents in TypeScript, facilitating the integration of various media types. For investors, this SDK represents an opportunity to support projects that leverage advanced AI capabilities, potentially leading to innovative applications and increased market competitiveness.

GLM 5.2 Fast via Wafer is now available on AI Gateway, achieving 2x higher throughput than competitors in both small and large contexts. It supports over 170 tok/s for small context and 200 tok/s for large context, with no platform fees on inference and a unified API for model management.
The release of GLM 5.2 Fast via Wafer on AI Gateway, which offers 2x higher throughput than competitors, is significant for builders and PMs as it allows for more efficient model deployment and management without platform fees. This could lead to reduced operational costs and faster iteration cycles, making it an attractive option for investors looking for scalable AI solutions.

The create-chat-sdk CLI simplifies bot development by scaffolding a complete Next.js project with platform and state adapters in one command. Users can easily select their desired integrations, such as Slack and Discord with Redis, and automate the setup process, making it ideal for both developers and CI environments.
The launch of the create-chat-sdk CLI streamlines the bot development process by enabling developers to quickly scaffold Next.js projects with necessary integrations like Slack and Discord. This reduces setup time and complexity, allowing builders and PMs to focus on enhancing user experience and functionality, while investors can see potential for faster market entry and reduced development costs.

Vercel now integrates with Claude Design, allowing users to deploy designs directly to Vercel for instant live URLs. This streamlines the process of sharing designs by connecting the two platforms seamlessly.
The integration of Claude Design with Vercel allows builders and PMs to deploy designs directly to live environments, significantly speeding up the feedback loop and iteration process. For investors, this development signals a growing trend towards seamless collaboration tools that enhance productivity in design and development workflows.

Chat SDK has integrated Kapso support via a new .vendor-official adapter, enabling seamless WhatsApp connectivity for bots. This integration simplifies WhatsApp Business setup, allowing developers to focus on bot logic while managing conversations, messages, and media through standard Chat SDK APIs.
The integration of Kapso support in Chat SDK via a new adapter streamlines WhatsApp Business setup for developers, allowing them to concentrate on bot functionality rather than infrastructure. This development enhances the attractiveness of the Chat SDK for builders and PMs looking to deploy effective messaging solutions quickly, while investors can recognize the potential for increased adoption and revenue growth in the bot development space.

The Chat SDK now integrates Linq support through the new .vendor-official adapter, enabling bots to handle texts in direct messages and group chats with full media and iMessage reaction capabilities. This integration allows seamless mapping of Linq chats to Chat SDK threads and standard APIs, ensuring consistent functionality across platforms.
The integration of Linq support into the Chat SDK through the .vendor-official adapter allows builders to enhance their bots with advanced messaging capabilities, including media handling and iMessage reactions. For PMs and investors, this development signals an opportunity to improve user engagement and streamline cross-platform communication, which can drive adoption and revenue growth.

Chat SDK now integrates Sendblue via the .vendor-official adapter, enabling bots to send and receive iMessage, SMS, and RCS messages. This integration provides features like iMessage-first delivery, automatic SMS and RCS fallback, and delivery status callbacks, enhancing user engagement across messaging platforms.
The integration of Sendblue with Chat SDK allows builders and PMs to enhance their messaging solutions by leveraging iMessage, SMS, and RCS capabilities, thus improving user engagement and communication efficiency. For investors, this development signals a growing trend towards multi-channel messaging solutions, which can drive user retention and monetization opportunities in the competitive messaging landscape.

Vercel AI has increased the project limit for hobby users from 10 to 25 projects per Git repository. This enhancement facilitates easier onboarding for monorepos and scenarios where a single codebase supports multiple deployable applications.
Vercel AI's increase of the project limit for hobby users from 10 to 25 projects per Git repository allows builders to manage more applications within a single codebase, streamlining development and deployment processes. For PMs and investors, this enhancement signals Vercel's commitment to supporting complex architectures, potentially leading to increased user engagement and retention.

Sakana AI's Fugu Ultra is now available on AI Gateway, utilizing a pool of frontier models for enhanced performance. It matches capabilities of Claude Mythos Preview and Fable 5, offering a unified API for model interaction without markup fees.
The availability of Sakana AI's Fugu Ultra on AI Gateway provides builders and PMs with access to a unified API that simplifies model integration, enhancing performance without markup fees. This development signals a competitive shift in the AI landscape, potentially lowering costs and increasing efficiency for developers and investors looking for scalable AI solutions.

The Agent Stack by Vercel AI provides essential building blocks for creating production-grade agents, enabling seamless integration across multiple AI models and secure operations. It features components like AI Gateway for model routing, Workflow SDK for durable execution, and Vercel Connect for scoped access, streamlining agent development and deployment across various platforms.
The Agent Stack by Vercel AI offers essential tools for building production-grade AI agents, which simplifies the integration of multiple AI models and enhances security. This development is significant for builders and PMs as it accelerates the deployment process, while investors should note its potential to streamline operations and reduce time-to-market for AI applications.
Vercel AI has eliminated CLI-specific deployment limits, enabling teams and AI agents to deploy from local machines and CI/CD pipelines without restrictions. This change allows for instant feedback and aligns deployment speed with workflow demands, enhancing overall efficiency.
Vercel AI's removal of CLI deployment limits allows teams to deploy without restrictions, significantly enhancing deployment speed and efficiency. This development is crucial for builders and PMs as it streamlines workflows, enabling faster iterations and more responsive development cycles, which can lead to improved product delivery and competitiveness in the market.

Vercel introduces its Enterprise Apps and Agents platform, enabling secure AI-driven app deployment with built-in access control and identity management through Vercel Passport, Connect, and Enterprise Managed Users, all while allowing integration with existing identity providers like Okta and AWS for enhanced security.
Vercel's launch of the Enterprise Apps and Agents platform, which includes Vercel Passport and integration with identity providers like Okta, enables builders and PMs to deploy secure AI-driven applications more efficiently. This development signals a growing focus on security and identity management in AI solutions, which is crucial for investors looking to support scalable and secure tech innovations.

The Workflow SDK 5 beta from Vercel AI introduces inflight cancellation support, allowing developers to manage operations across workflows and steps using AbortController and AbortSignal. This feature enables cancellation of ongoing tasks, enhances cooperation between steps, and maintains signal durability across suspensions, improving overall workflow efficiency.
The introduction of inflight cancellation support in Vercel AI's Workflow SDK 5 beta allows developers to efficiently manage and cancel ongoing tasks, enhancing workflow efficiency and responsiveness. This capability can significantly reduce resource wastage and improve user experience, making it a valuable tool for builders and PMs focused on optimizing application performance.

Access to Claude Fable 5 has been suspended for all users on the AI Gateway due to compliance with a US Government directive. There is currently no information on when or if access will be restored, but users can still utilize other Anthropic models available on the platform.
The suspension of Claude Fable 5 access on the AI Gateway due to a US Government directive highlights the regulatory risks associated with AI models, which could impact project timelines and resource allocation for builders and PMs. Investors should be aware of these compliance challenges as they may affect the viability and scalability of AI solutions in the market.

GLM 5.2 is now available on AI Gateway, featuring an upgraded context window of 1M tokens, significantly improving reliability for long-horizon tasks. The model adheres to engineering standards and is accessible via a unified API without markup or platform fees on inference, including BYOK requests.
The release of GLM 5.2 on AI Gateway, featuring a 1M token context window, enhances the model's reliability for long-horizon tasks, which is crucial for builders and PMs developing applications that require handling extensive data. This upgrade also eliminates markup and platform fees, making it more cost-effective for investors looking to support scalable AI solutions.

Vercel AI introduces a unified API for agent harnesses like Claude Code, Codex, and Pi, allowing seamless model switching without code changes. The AI SDK 7 enables developers to manage components such as skills and sessions through a single abstraction, enhancing safety with sandboxed environments.
Vercel AI's introduction of a unified API for agent harnesses like Claude Code, Codex, and Pi simplifies the integration of multiple AI models, allowing developers to switch between them without altering code. This enhances development efficiency and safety, making it easier for builders and PMs to create robust applications while also appealing to investors looking for scalable solutions in AI development.

The Kimi K2.7 Code from Moonshot AI is now available on AI Gateway, designed for long-horizon programming tasks across various scenarios. It features a multimodal architecture supporting both text and vision inputs, and operates in thinking mode for enhanced performance.
The release of Kimi K2.7 Code on AI Gateway introduces a multimodal architecture that enhances long-horizon programming tasks by integrating text and vision inputs. This development allows builders and PMs to create more sophisticated applications, while investors can recognize the potential for improved efficiency and innovation in AI-driven solutions.

Okara leverages Vercel AI to manage marketing for over 120,000 businesses, processing 4 billion tokens daily. The AI CMO automates SEO, content, and social media strategies, enabling solo founders to achieve the marketing impact of larger teams without the high costs of traditional agencies.
Okara's use of Vercel AI to automate marketing for over 120,000 businesses demonstrates a scalable solution that empowers solo founders to compete effectively without large budgets. This development signals a shift towards AI-driven marketing strategies, highlighting opportunities for builders and investors to explore cost-effective tools that enhance productivity and reduce reliance on traditional marketing agencies.

Okara leverages Vercel AI to manage marketing for over 120,000 companies, processing 4 billion tokens daily with a team of just four. Their AI CMO automates SEO, content, and social media strategies, enabling solo founders to achieve distribution equivalent to larger teams without hefty agency costs.
Okara's use of Vercel AI to automate marketing for 120,000 companies with just four team members demonstrates the potential for significant cost savings and efficiency in marketing operations. Builders and PMs should consider how similar AI-driven solutions can scale their offerings, while investors may see opportunities in technologies that enable small teams to compete with larger agencies.

DeepSeek V4 Pro and V4 Flash models are now accessible via Azure on AI Gateway, enabling automatic routing and failover without code changes. Users can prioritize Azure requests and utilize their own Azure credentials for enhanced API key authentication.
The availability of DeepSeek V4 Pro and V4 Flash models on Azure's AI Gateway allows builders and PMs to integrate advanced AI capabilities with minimal code changes, enhancing operational efficiency. For investors, this signifies a growing trend towards seamless AI deployment in cloud environments, potentially increasing the value of companies leveraging these technologies.
Vercel AI has introduced threshold billing for Pro teams, allowing partial invoices to be sent mid-cycle when on-demand usage hits a specified threshold. This change ensures that users are billed for their usage incrementally rather than waiting until the end of the billing period, while preventing double billing for the same usage.
Vercel AI's introduction of threshold billing for Pro teams allows for incremental invoicing based on usage, which enhances cash flow management for builders and PMs. This development reduces the risk of unexpected charges at the end of the billing cycle, making it easier for teams to budget and plan their expenditures effectively.

Claude Fable 5 from Anthropic is now available on AI Gateway, enhancing performance on complex tasks with improved first-shot correctness and reduced human intervention. The model features built-in classifiers to mitigate misuse risks and retains prompts for 30 days without zero data retention support. AI Gateway offers a unified API for model management without platform fees.
The release of Claude Fable 5 on AI Gateway is significant for builders and PMs as it enhances performance on complex tasks, allowing for more efficient product development with reduced human oversight. For investors, the introduction of built-in classifiers to mitigate misuse risks signifies a more responsible approach to AI deployment, potentially increasing market confidence and adoption rates.