Service eligibility and regional restrictions
PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.
Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.
Key Takeaways
- The offer covers Parallel web search in fast mode on AI Gateway.
- Use standard Gateway authentication. No separate Parallel API key is required.
- If you changed eve's search provider or use a direct model provider, check the actual tool route.
- The offer enables Zero Data Retention by default within its documented scope.
Identify the product and mode you are using
The announcement covers the Parallel web search tool in fast mode through Vercel AI Gateway. Search returns links and excerpts, with options to filter by site or publication date. Vercel's model catalog example uses basic. If you copy it, check the mode before running searches to avoid standard rates.
Parallel Decisions Fast is a separate product for structured decisions with evidence, using the model ID parallel/parallel-decisions-fast. Its own offer runs through November 23, 2026. When reviewing a request, record the generation model, search tool, and search mode separately.
Select fast in the AI SDK and check eve's provider
Register gateway.tools.parallelSearch({ mode: 'fast' }) under tools in the AI SDK generation call, and use an AI Gateway model. Authenticate to Gateway as usual. Vercel says a separate Parallel API key is unnecessary. Start with a prompt that clearly needs recent web information and confirm the run actually called the search tool.
eve uses Parallel fast automatically when a model is served through AI Gateway and the default built-in web_search remains in place. If you selected another search provider, replaced the built-in tool, or connect directly to a model provider, check your actual configuration. eve documents native search for direct providers, so review charges for the route you used.
Check search usage and model charges separately
At the time of review, Vercel's standard catalog lists $5 per 1,000 searches and $1 per 1,000 additional results beyond 10 per search. The November 18 announcement applies the offer to fast mode and says other modes are billed at standard rates. In usage reporting, distinguish promotional search requests, other search modes, and the generation model's input and output.
The announcement says the offer runs through January 31, 2027. It gives no cutoff time or time zone. Recheck the offer and your usage reporting as the date approaches. Vercel also says Zero Data Retention is on by default for this offer. Review the retention rules for the model and tool and the handling of your application's own logs separately.
Sources
Frequently Asked Questions
Do I get the fast-mode offer if I omit mode?
Set mode: 'fast' explicitly in the AI SDK. Vercel lists basic as the balanced default. The built-in eve default uses fast automatically for models served through AI Gateway.
Does the search offer also make model generation free?
The offer covers the search tool. Check generation usage against the chosen model's rates and any applicable credits.
Can I call the tool without a separate Parallel key?
Through AI Gateway, no separate Parallel API key is required. Standard Gateway authentication still applies. Check the authentication rules for other entry points separately.