AI Tools Review

Gemini 4 Carbon

By Google

Released: 2026-10-28

LLM
Google
Paid
New

Gemini 4 Carbon is Google's AI model, released 28/10/2026. Gemini 4 Carbon is Google DeepMind's top Gemini 4 model, based on Gemini 4 Argon and aimed at long-horizon agentic work, cybersecurity and scientific reasoning. The vendor's table has it ahead of Argon, GPT-6 Astra, Claude Opus 5.5 and Claude Fable 5.1 on most rows, with Astra ahead on FrontierSWE 2 and Opus 5.5 ahead on OSWorld 2.1. Cybersecurity results reflect access through the Fairwind Program for verified defenders.

Visit Gemini 4 Carbon
Read our full in-depth review: Gemini 4 Carbon: Is It Worth the Price?

AI-Powered

Leverages advanced AI technology to deliver cutting-edge capabilities and results.

Fast & Efficient

Optimized performance ensures quick results without compromising on quality.

Purpose-Built

Specifically designed for llms tasks and workflows.

How does Gemini 4 Carbon score on independent benchmarks?

Gemini 4 Carbon ranks #4 of 28 models on the Artificial Analysis Intelligence Index (v5.0.1) with a score of 60, as measured on 22 November 2026.

Intelligence Index
60
AA v5.0.1
Coding Agent Index
55.5
in Antigravity CLI
Cost per task
$7.76
AA Intelligence Index task
Price per 1M tokens
$4 / $20
input / output

Where does Gemini 4 Carbon rank against other models?

How good is Gemini 4 Carbon at agentic coding?

How does Gemini 4 Carbon do on individual tests?

Evaluation (run by Artificial Analysis)Gemini 4 CarbonBest of any model
Terminal-Bench 4.065.2%67.2%
FrontierCode 2.040.6%48.8%
Humanity's Last Examn/a74.3%
GDPval-AA (Elo)17331866

How has Gemini 4 Carbon's score changed over time?

We have tracked Gemini 4 Carbon on Intelligence Index v5.0.1 since 22/11/2026. A point is added each day the score, coding index or price changes, so this chart fills in over time.

What does the Gemini 4 Carbon system card report?

Gemini 4 Carbon is Google DeepMind's top Gemini 4 model, based on Gemini 4 Argon and aimed at long-horizon agentic work, cybersecurity and scientific reasoning. The vendor's table has it ahead of Argon, GPT-6 Astra, Claude Opus 5.5 and Claude Fable 5.1 on most rows, with Astra ahead on FrontierSWE 2 and Opus 5.5 ahead on OSWorld 2.1. Cybersecurity results reflect access through the Fairwind Program for verified defenders.

These are Google DeepMind's own published results (self-reported, with the vendor's harness and settings), so they are not directly comparable with other labs' numbers. Sources: Gemini 4 Carbon Model Card (undefined/11/2026), Gemini 4 Carbon Model Evaluation (approach, methodology and results) (undefined/10/2026), Introducing Gemini 4 Carbon (28/10/2026). Last checked 22/11/2026.

Other

BenchmarkGemini 4 CarbonGemini 4 ArgonGPT-6 AstraClaude Opus 5.5
Artificial Analysis Intelligence Index v5.0with tools; highest thinking59 points505758
Vals Index v2.2Vals AI sourced70.4%69.2%65%70.2%
CWE-bench v1 (cybersecurity)programmatic pass@1; accessed through the Fairwind Program71%68%68%67%
ExploitGym (cybersecurity)Antigravity harness, mitigations off, single attempt; 6 hour default and 10 hour extended timeouts reported in methodology; Fairwind Program54.8%42%42.4%34.5%
SEC-bench Pro (cybersecurity)pass@1 on 183-task May 2026 JavaScript engine split, 3 attempts, Antigravity harness; Fairwind Program88.2%83.4%85.4%n/a
WebDev Arenarank #4, up 89 points per launch post1768 Elon/an/an/a

Coding

BenchmarkGemini 4 CarbonGemini 4 ArgonGPT-6 AstraClaude Opus 5.5
SWE-sweepsubtask-weighted reward; mini-swe-agent harness; self-computed9.6%4.3%9.5%3.6%
FrontierSWE 2Proximal official public leaderboard61.2%55%65.5%62.3%
Terminal-Bench 4.0 (AA)Artificial Analysis sourced; cheating reduced to 0.0% per launch post65.2%57.1%59.1%59.6%

Agents & tool use

BenchmarkGemini 4 CarbonGemini 4 ArgonGPT-6 AstraClaude Opus 5.5
Agents' Last Exam v1.1score; 122-task set, ALE-Claw harness, 5-hour window, safety filters on; self-computed77.4%69%73.1%71.6%
OSWorld 2.182-task offline subset, max of 3 runs, partial score; self-computed81.5%74.7%80.9%87.2%
AutomationBenchZapier private set, official leaderboard53.9%51.3%41.4%42.5%
Vending-Bench 2final bank balance; Andon Labs leaderboard16484 USD13718155159235

Multimodal

BenchmarkGemini 4 CarbonGemini 4 ArgonGPT-6 AstraClaude Opus 5.5
Chartographyno tools; Surge leaderboard72.5%71.6%71%66.3%
Blueprint-Bench 2similarity score; Andon Labs leaderboard58.6%54.4%49.7%51.2%

Long context

BenchmarkGemini 4 CarbonGemini 4 ArgonGPT-6 AstraClaude Opus 5.5
GraphWalks (256k to 1M)BFS F1 on 200 problems between 256k and 1M tokens; self-computed96.2%84.2%71.8%66.8%

Reasoning & maths

BenchmarkGemini 4 CarbonGemini 4 ArgonGPT-6 AstraClaude Opus 5.5
RiemannBenchSurge public leaderboard80%76%72%69.6%

Knowledge & science

BenchmarkGemini 4 CarbonGemini 4 ArgonGPT-6 AstraClaude Opus 5.5
Terminal-Bench Science 0.1 (AA)Artificial Analysis sourced64.8%52.9%63.3%59%

Safety & honesty

BenchmarkGemini 4 CarbonGemini 4 ArgonGPT-6 AstraClaude Opus 5.5
Gray Swan IPI (attack success rate, lower is better)indirect prompt injection; per launch post0%0.7%n/an/a

What did safety testing find for Gemini 4 Carbon?

Safety level: Frontier Safety Framework (April 2026): full assessment is in the separate Gemini 4 Carbon Safety and Security Report; Google could not rule out that Argon and Carbon reached the biological CBRN CCL and the Cybersecurity CCL

  • Versus Gemini 4 Argon, automated safety evaluations: text-to-text +3.50pp, multilingual +3.55pp, image-to-text -1.00pp, tone +0.10pp, unjustified refusals +0.10pp.
  • Human red teaming: child safety evaluations met required launch thresholds, with similar or improved content safety versus Argon and no egregious concerns found.
  • Google strengthened Frontier Safety safeguards for CBRN and cybersecurity misuse and provides specialised dual-use cyber capabilities only to verified defenders through the Fairwind Program.
  • Gray Swan indirect prompt injection attack success is 0.0% versus 0.7% for Argon, per the launch post.
  • Google says it is participating in the US government's voluntary pre-release model testing process.

Independent figures: Artificial Analysis, refreshed daily on our AI model benchmarks page. Full analysis: our Gemini 4 Carbon deep dive.

Google Model Timeline

Gemini 4 Neon

1M tokens (up to 64K default synchronous output, up to 1M via background long decode) context

Gemini 4 Hydrogen

1M tokens (up to 64K default synchronous output, up to 1M via background long decode) context

Gemini 4 CarbonCurrent

1M tokens (combined input and output) context

Google Omni
Google Spark
Gemini 3.5 Flash

1M tokens context

Google: Gemini 3 Flash Preview

1,049k tokens context

Google: Gemini 3 Pro Preview

1,049k tokens context

Google: Gemini 2.5 Flash Preview 09-2025

1,049k tokens context

Google: Gemini 2.5 Flash Lite Preview 09-2025

1,049k tokens context

Google: Gemini 2.5 Flash Lite

1,049k tokens context

Google: Gemma 3n 2B (free)

8k tokens context

Google: Gemini 2.5 Flash

1,049k tokens context

Google: Gemini 2.5 Pro

1,049k tokens context

Google: Gemini 2.5 Pro Preview 06-05

1,049k tokens context

Google: Gemma 3n 4B (free)

8k tokens context

Google: Gemma 3n 4B

33k tokens context

Google: Gemini 2.5 Pro Preview 05-06

1,049k tokens context

Google: Gemma 3 4B (free)

33k tokens context

Google: Gemma 3 4B

96k tokens context

Google: Gemma 3 12B (free)

33k tokens context

Google: Gemma 3 12B

131k tokens context

Google: Gemma 3 27B (free)

131k tokens context

Google: Gemma 3 27B

96k tokens context

Google: Gemini 2.0 Flash Lite

1,049k tokens context

Google: Gemini 2.0 Flash

1,049k tokens context

Google: Gemini 2.0 Flash Experimental (free)

1,049k tokens context

Google: Gemma 2 27B

8k tokens context

Google: Gemma 2 9B

8k tokens context

Specifications

pricing$4/$20 per 1M tokens (input/output)
context Window1M tokens (combined input and output)