Gemini 4 Carbon lands on AI Gateway: a 1M-token context and half-price calls through November 30, 2026
Synopsis
Google's Gemini 4 Carbon is now available on AI Gateway under the model name google/gemini-4-carbon, aimed at complex software engineering, sustained agentic work and cybersecurity defense, with a 1M-token context window for large codebases and long inputs. Google reports improvements in coding, full-stack web development and scientific reasoning on multi-stage engineering tasks, cybersecurity coverage from finding vulnerabilities through validating risk to developing patches, and more reliable multi-step plan execution, fewer unsupported claims and better collaboration when requirements are ambiguous. Calls through AI Gateway reflect provider pricing with no markup and no platform fee on inference, at 50% off through November 30, 2026.
Interpretation
Google's Gemini 4 Carbon can now be called on AI Gateway as google/gemini-4-carbon, positioned for complex software engineering, sustained agentic work and cybersecurity defense. Previously this work required reaching the model elsewhere; it now appears as a selectable model on AI Gateway, paired with a 1M-token context window for large codebases and long inputs. Availability comes from the AI Gateway release note, which includes a runnable call example and setup steps for coding agents such as Claude Code, Codex and Cursor.
The model carries engineering tasks through multiple stages: Google reports improvements in coding, full-stack web development and scientific reasoning, and cybersecurity coverage from finding vulnerabilities to validating risk to developing patches. Google also reports more reliable execution of multi-step plans, fewer unsupported claims and better collaboration when requirements are ambiguous, so agents follow through while respecting project constraints. These are Google's reported capability descriptions; the release note gives no benchmark scores, test sets or comparison figures.
Access stays simple: install the latest Vercel CLI, run npx vercel ai-gateway setup, then select google/gemini-4-carbon in the agent. AI Gateway itself provides a unified API for calling models, tracking usage and cost, and configuring retries, failover and performance optimizations. Billing reflects provider pricing with no markup and no platform fee on inference, including Bring Your Own Key (BYOK) requests; calls are 50% off through November 30, 2026.
Perspective
The report is limited to the AI Gateway release note: model availability, call name, 1M-token context, Google's reported capability directions, setup steps and billing. The intended readers are researchers and engineers integrating the model into coding agents or applications.
The improvements in coding, full-stack web development, scientific reasoning and cybersecurity are Google's reports; the release note gives no benchmark scores, test conditions or comparisons with other models, and the claims of more reliable multi-step execution, fewer unsupported claims and better collaboration under ambiguous requirements are likewise unquantified.