DeepSeek V4.1 Flash is now available in Token Pass

Vercel's Token Pass, a flat monthly plan for running coding agents on a short list of models, now includes DeepSeek's V4.1 Flash alongside Moonshot AI's Kimi K3 and Z.ai's GLM 5.3. The plan costs Pro teams $30 a month and gives $80 of monthly credit spent through Vercel's AI Gateway, the service that routes requests to model providers, with tools such as Cline and Pi or custom agents built on the AI SDK. The announcement says the DeepSeek model runs on its latest 1017 weights.

Two details shape what the bundle means in practice. Vercel's launch post for Token Pass says unused credit does not roll over and that requests stop when the credit runs out until the next billing period, with no automatic overage charges, so the plan works as a capped budget rather than open-ended usage. Zero Data Retention is on by default, meaning the gateway and provider delete prompts and responses after each request. The 1017 snapshot, as described in Vercel's Gateway listing, is a minor update to V4.1 Flash that the company says improves reasoning, coding, agentic tasks and image understanding and finishes tasks with shorter outputs; this short changelog adds no benchmarks of its own.