Mistral Large 4 updated weights and lower pricing on AI Gateway
By Steven Van ·
Existing mistral/mistral-large-4 calls on AI Gateway pick up the new weights and lower token prices automatically, with no change to the model ID.
Mistral Large 4 on Vercel's AI Gateway now runs on updated weights, and input, cached input and output tokens all cost less. Mistral's release announcement says the new weights improve agentic coding and defensive ccybersecurity, with greater token efficiency on coding tasks. The release also improves spreadsheet creation and editing, web research, and resistance to prompt injection.
No change needed for existing users
Anyone already using mistral/mistral-large-4 through Vercel AI Gateway does not need to change their model. The existing model ID string now points to the updated weights and the new pricing.
Using it in coding agents
To use the model in fx, Pi, Claude Code, Codex, Cursor or other harnesses, install the latest Vercel CLI and run npx vercel ai-gateway setup. Then select mistral/mistral-large-4 in the agent. The coding agents guide covers the setup in detail. According to the guide, the command detects the agents installed on a machine, shows a diff of every planned change before writing anything, and stores the key in the macOS Keychain rather than in plaintext config.
The model can also be tried in the model playground. AI Gateway reflects provider pricing with no markup and charges no platform fee on inference, including on Bring Your Own Key requests.