Mistral Large 4 gets updated weights and lower prices on Vercel AI Gateway
Mistral Large 4 on Vercel's AI Gateway now runs on updated weights and costs less for input, cached input and output tokens. Developers already using the model do not need to change anything to get the new version.
- Mistral Large 4 on Vercel AI Gateway now uses updated weights.
- Pricing is lower for input, cached input and output tokens; exact prices are not given in the source.
- According to Mistral, the update improves agentic coding, defensive cybersecurity, spreadsheet work, web research and prompt injection resistance.
- The existing model ID mistral/mistral-large-4 switches to the new version automatically.
- Vercel says AI Gateway applies no markup and no platform fee on inference, including BYOK requests.
What changed in Mistral Large 4
Vercel says Mistral Large 4 from Mistral now uses updated weights on AI Gateway, and that pricing is lower across three token categories: input, cached input and output. The changelog does not list the new prices, so the size of the reduction is not stated in the source.
The capability changes come from Mistral. According to Mistral's release announcement, as relayed by Vercel, the updated weights improve agentic coding and defensive cybersecurity, with greater token efficiency on coding tasks. The release also improves spreadsheet creation and editing, web research, and resistance to prompt injection.
No migration needed
Teams that already call mistral/mistral-large-4 through AI Gateway do not have to switch models to use the update. According to Vercel, the existing model ID string uses the new model with updated weights and pricing.
Vercel's example uses the AI SDK. It calls streamText with the model set to 'mistral/mistral-large-4' and a prompt asking for three fixes for a flaky API retry loop, then writes the streamed text to standard output.
For coding agents and similar tools, Vercel lists fx, Pi, Claude Code, Codex and Cursor, among other harnesses. Users should install the latest Vercel CLI and run npx vercel ai-gateway setup, then select mistral/mistral-large-4 in the agent. Vercel points to its coding agents guide for details, and the model can also be tried in the model playground.
About AI Gateway and what it means for users
Vercel describes AI Gateway as a unified API for calling models, tracking usage and cost, and configuring retries, failover and performance optimizations for higher-than-provider uptime. Built-in features include custom reporting, Zero Data Retention support, budgets for API keys and routing rules.
On pricing, Vercel states that AI Gateway reflects provider pricing with no markup and does not charge a platform fee on inference, including on Bring Your Own Key (BYOK) requests. That means the lower Mistral Large 4 rates apply directly to users of the gateway, according to the company.
For developers, the practical effect is an automatic switch to newer weights at lower token costs, with the stated gains focused on coding, cybersecurity, spreadsheet work, web research and prompt injection resistance. Vercel's changelog does not include benchmark figures, so teams that depend on those areas will need to test the updated model on their own workloads.
