Latest in AI
Ling 3.1 Flash 1019 on Vercel AI Gateway: Free Access, Context and API Details
The updated Ling model reaches AI Gateway with a larger context window and a short free period that will turn into paid usage.
Vercel added InclusionAI's Ling 3.1 Flash 1019 to AI Gateway on October 30, with a 1M-token context window and free access until November 3. The new Gateway listing gives developers a short evaluation window for coding and tool-using agents, but calls will begin billing when the promotion ends.
Ling 3.1 Flash 1019 pricing and availability
The official announcement says requests start charging at the published rate on November 3, 2026. It does not specify an exact cutoff time or timezone, so do not assume the offer lasts through the end of that day in your location.
At publication, the model catalog lists Novita AI as the provider and marks input and output pricing free. The checked listing does not establish the post-promotion dollar rate. Check the live catalog before leaving a test workload running beyond the offer.
| Gateway detail | Current listing |
|---|---|
| Model ID | inclusionai/ling-3.1-flash-1019 |
| Context window | 1M tokens |
| Maximum output | 262,144 tokens |
| Provider listed | Novita AI |
| Promotional pricing | Free until November 3 |
The context window is shared by prompt and response, and reasoning tokens count toward the output limit. A 1M-token context window is not a 1M-token input allowance plus an independent output budget.
What changed from the earlier Gateway listing?
Vercel's September 30 Ling 3.1 Flash announcement listed a 262K-token context window and an earlier free period ending in October. The October 30 story is the updated 1019 version arriving on Gateway, not Ling's first appearance there.
That earlier release also documented a separate -free model ID which would stop serving when its promotion ended. The new announcement instead gives the 1019 ID and explicitly says requests will begin charging. Do not carry the old promotion's stopping behavior over to the new identifier.
How to use the API
Applications using the AI SDK can select inclusionai/ling-3.1-flash-1019 with generateText or streamText. The catalog also documents access through OpenAI Chat Completions, OpenAI Responses and Anthropic Messages-compatible interfaces. This is a provider integration, not a requirement to host the model's weights yourself.
For coding agents, Vercel's announcement directs developers to run npx vercel ai-gateway setup using the latest CLI, then select the model.
Vercel's API-key documentation also allows budgets on individual keys. A separate evaluation key with a deliberate spending limit is easier to manage than a shared production credential when a temporary offer changes to paid use.
What the free period can actually tell you
A larger accepted context does not establish that the model will find every important detail in a large repository or reliably finish a long agent task. Those are separate properties to test. The Gateway announcement supplies availability and integration details, not an independent benchmark comparison against your existing model.
Use the promotion to test a bounded workflow, not to assume permanent economics. Keep the tasks, tool permissions and success criteria fixed, then compare completed work and failure modes. The useful result is knowing whether the model earns a place in that workflow after the free period ends.
I'm Paras. I work at the intersection of AI, engineering and distribution.
I help technical founders turn model launches and complex product updates into clear content developers and founders can understand and share.