Gemini 4 Carbon: price, API access and my own test

Update 10 November 2026: Gemini 4 Carbon is available to everyone

Since 10 November 2026 Gemini 4 Carbon is generally available. Model ID gemini-4-carbon in the Gemini API and Google AI Studio (paid tier only, no free tier), google/gemini-4-carbon in OpenRouter, and Google AI Ultra in the Gemini app, AI Mode and Google Antigravity (Low, Medium and High thinking levels). It is not part of Google AI Pro.

Context window: 1,048,576 tokens shared by input and output, up to 65,536 output tokens by default. Google AI Ultra: .99 (5x) or 9.99 (20x) a month.

My first test (10 Nov): 23 practical tasks all correct (0.23 USD, 264 s - Argon took 135 s), 16 of 16 scientific sources real with 0 invented, 8 of 8 planted vulnerabilities found in a code review, a 639-line spreadsheet engine passing 34 of 34 tests. Weak spot: at medium effort it twice used its whole token limit on thinking and returned no answer. Details: Gemini 4 Carbon is available and my Gemini 4 Carbon test vs Argon, Claude Fable 5.2 and GPT-6.1 Sol. The sections below were written on 28 October, before public access; prices and access above are current.

Access and release date

The announcement is not a public release. Google says selected cyber defenders begin getting access this week, followed by trusted partners through Fairwind. Developers, businesses and consumers are to follow “as soon as possible”, without a calendar date. Paying for a Gemini subscription does not establish access to Carbon.

Fairwind is aimed at governments, national cyber bodies, critical infrastructure operators, major technology platforms and academic laboratories. Google describes more than 650 partners, but does not name eligible countries. The Fairwind page still describes Argon. A US developer or security team should treat the interest form as an application, not an instant API-key request.

API pricing and US subscriptions

Google lists introductory API rates of USD 2 per million input tokens and USD 10 per million output tokens, then USD 4 and USD 20. Cached input is 95% cheaper than standard input. There is no published end date for Carbon’s introductory period. Argon’s USD 2/10 offer ends on 30 November 2026; that date must not be copied to Carbon. These are announced rates, not evidence that the public endpoint is open.

US monthly Google AI subscription prices are context for Argon, not a Carbon access offer. Pro and Ultra list Argon; Plus is not the Argon tier. Subscription fees and metered API usage are different purchases. Prices below were read on Google One on 28 October 2026; they are not a conversion into other English-speaking markets.

Google One - 28.10.2026

Google-reported benchmarks

Google’s report uses pass@1 and the highest thinking settings through the Gemini API. It runs SWE-sweep itself with mini-swe-agent, and ExploitGym and SEC-bench Pro with the Antigravity harness, without internet and with mitigations disabled. Google also ran Agents’ Last Exam and OSWorld internally. An external benchmark name does not make every submitted score an independent replication.

Coding and agents

Reasoning and visual tasks

Cybersecurity

According to Google, Carbon reaches 88.2% on SEC-bench Pro for vulnerability detection, 54.8% on ExploitGym for exploitation and 71.0% on CWE-bench for patching. These evaluations were performed without guardrails. They describe capability under those conditions, not a safe deployment policy or permission to run an agent against production systems.

Honesty and agent behaviour

Google reports successful cheating at 0.0% and cheating attempts at 0.4% on Terminal-Bench 4.0. It also reports nearly 80% fewer serious misalignment alerts in live coding, about half as many unsupported claims as Argon, 26% less overreach and 29% less under-service. Gray Swan IPI successful prompt injections fall from Argon’s 0.7% to 0.0%. These are bounded evaluations, not a guarantee that an agent cannot be misled.

Google says Carbon worked for days on internal investigations into subtle bugs in large codebases and low-level systems. It also claims more than twice Argon’s effectiveness at generating secure code. Chrome’s reported gain of over 10 times concerns high-quality patches against an unnamed frontier competitor. Neither claim gives us an independent measure of time saved on a commercial repository.

Carbon vs Argon

The useful Carbon vs Argon question is whether the improvement matches the job. Google reports a rise from 57.1% to 65.2% on Terminal-Bench and from 84.2% to 96.2% on GraphWalks. Yet Carbon is not the best model in every row: GPT-6 Astra scores 65.5% against 61.2% on FrontierSWE 2, while Opus 5.5 reaches 87.2% against 81.5% on OSWorld. Those competing scores are reproduced from Google’s table.

What the results do not establish

There is no English writing test, latency measurement or independent cost-per-completed-task result here. The announced token price cannot tell you how much an agent will spend over several days. A ranking index is also not a percentage of all work completed correctly. Before changing a production workflow, wait for access and compare the same repository, acceptance checks and spending cap. That is a proposed evaluation, not one we have run.

Related reading

Existing Gemini 4 Argon guide | Gemini 4 Carbon

Gemini 4 Carbon benchmarks: what Google’s scores show

Sources - accessed 28 October 2026

• Google - Introducing Gemini 4 Carbon - 28.10.2026
• Google, Gemini 4 Carbon model evaluation - 28.10.2026
• Fairwind - 28.10.2026
• Sundar Pichai (X) - 28.10.2026
• Google (X) - 28.10.2026
• Gemini 4 Argon safety report - 28.10.2026
• Google One - 28.10.2026