inclusionAI's Ling 3.0 Flash Sante runs free on Vercel AI Gateway through October 4

Vercel has added Ling 3.0 Flash Sante, a health and medicine-focused large language model from inclusionAI, to its AI Gateway and is making it free to use through October 4, according to the Vercel blog. The move gives developers a low-friction way to evaluate a domain-specific medical model through an API they may already use, without signing up with another provider.

The model

Ling 3.0 Flash Sante is a medically tuned variant of Ling 3.0 Flash. Vercel describes it as a Mixture-of-Experts (MoE) model with 124 billion total parameters, of which roughly 5.1 billion are active per token, alongside a 256K token context window and support for function calling. Sparse activation at that scale typically means lower inference cost and latency than an equivalently sized dense model, which fits the "Flash" positioning of the base family.

According to the Vercel blog, the model is built for clinical reasoning and professional healthcare tasks, deep research, evidence-based retrieval, and multi-step medical workflows, while retaining the base model's general reasoning, coding and agentic capabilities. In practice, a single model can therefore handle both the medical prompts and the tool-calling scaffolding wrapped around them.

Two model IDs, two outcomes after the deadline

Vercel exposes the model under two identifiers, and the difference matters once the promotional window closes. The standard ID, inclusionai/ling-3.0-flash-sante, is free through October 4 and then begins billing at normal rates. The free ID, inclusionai/ling-3.0-flash-sante-free, ceases to serve requests when the promotion lapses rather than switching to paid billing — a safer option for prototypes that should stop working instead of quietly racking up charges.

Requests against the free model still appear in the spend dashboard and carry trace data, Vercel notes; they are metered at zero cost rather than excluded from reporting, so teams can observe usage exactly as they would for a paid model.

On the usage side, developers set the model ID in the AI SDK. For coding agents, Vercel points to its coding agents guide and the vercel ai-gateway coding-agents setup command, which connects tools such as Claude Code, Codex and Cursor before selecting the model inside the agent. A model playground and a dedicated page for the free variant are also available for trying it without writing code.

The gateway behind the offer

The promotion doubles as a showcase for AI Gateway, Vercel's unified inference layer. According to the Vercel blog, the gateway provides a single API for calling models across providers, tracks usage and cost, and handles retries, failover and performance optimizations that Vercel positions as delivering better uptime than calling providers directly. It also ships with custom reporting, Zero Data Retention support, budgets for API keys and routing rules. Pricing passes provider rates through with no markup and no platform fee on inference, including on Bring Your Own Key requests.

Why it matters

Domain-specific models are often harder to trial than general-purpose ones, since they tend to sit behind enterprise agreements or require self-hosting. Free API access to a 124B-parameter medical model — with traces, spend dashboards and function calling included — removes most of that friction for the next month.

The two-ID setup is a practical touch as well: teams can experiment with the free variant knowing it will shut down rather than bill, then move to the standard ID only if the model earns a place in production. For anyone building healthcare-adjacent applications, agents or research tooling, October 4 is the cutoff to evaluate Ling 3.0 Flash Sante at no cost.