← All models

Gemini 4 Carbon

Language Available Full comparison ↗

google-deepmind/gemini-4-carbon · by Google DeepMind · decoder-only-transformer

Pricing — 1 offering(s)

Google DeepMind

About this offering: Gemini 4 Carbon, Google's most capable reasoning model, generally available on the Gemini API since 2026-11-10 (paid tier only; not on the Free tier). It sits above Gemini 4 Argon at the same price, and uses more tokens per task to explore and verify. One price at every prompt length, with no >200K long-context band. Google's introductory pricing runs through November 30, 2026, with no timezone given; Modelglass shows the standard price from the November 30 rebuild, labelled from December 1.

Input tokens

  • $2.00 / 1M tokens (input) Current 2026-11-10 → 2026-11-30

    Introductory rate, $2.00/1M input through 2026-11-30; $4.00/1M from 2026-12-01 per Google's pricing page.

  • $4.00 / 1M tokens (input) Future 2026-12-01

Cached input tokens

  • $0.10 / 1M tokens (cached input) Current 2026-11-10 → 2026-11-30
    Implicit caching only (explicit caching is not supported); 95% off input. Introductory rate, $0.10/1M through…

    Implicit caching only (explicit caching is not supported); 95% off input. Introductory rate, $0.10/1M through 2026-11-30; $0.20/1M from 2026-12-01.

  • $0.20 / 1M tokens (cached input) Future 2026-12-01

Output tokens

  • $10.00 / 1M tokens (output) Current 2026-11-10 → 2026-11-30
    Includes thinking tokens. Introductory rate, $10.00/1M output through 2026-11-30; $20.00/1M from 2026-12-01…

    Includes thinking tokens. Introductory rate, $10.00/1M output through 2026-11-30; $20.00/1M from 2026-12-01 per Google's pricing page.

  • $20.00 / 1M tokens (output) Future 2026-12-01

Batch

Batch input tokens

  • $1.00 / 1M tokens (input) Current 2026-11-10 → 2026-11-30
  • $2.00 / 1M tokens (input) Future 2026-12-01

Batch output tokens

  • $5.00 / 1M tokens (output) Current 2026-11-10 → 2026-11-30
  • $10.00 / 1M tokens (output) Future 2026-12-01

Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.

See how Gemini 4 Carbon fits into a cost-aware routing setup

See how →

Capability profile

How Gemini 4 Carbon rates across core capability dimensions, with the task-level evidence behind each rating.

reasoning Strong

Google calls it its most capable reasoning model, for tasks that test the limits of a model's intelligence, with tunable thinking levels (low, medium, high).

coding Strong

Google's release notes: built for sustained agentic coding. Google reports a #1 rank on CursorBench 4.0 (vendor-reported). No independent coding benchmark recorded yet.

tool use Strong

Function calling, code execution, Google Search/Maps grounding and URL context are listed as supported on Google's model page. Google reports #1 ranks on Zapier's AutomationBench and Perplexity's WANDR (vendor-reported).

instruction following Strong
context window Strong

1,048,576-token unified input and output window; 65,536 output tokens by default, up to about 1M with long decode.

multilingual Strong
speed Weak

Google says Carbon uses more tokens than Gemini 4 Argon to explore and verify, so expect slower, longer responses than Argon.

cost efficiency Moderate

Same price as Gemini 4 Argon: $2/$10 per 1M introductory pricing through 2026-11-30, then $4/$20. It spends more tokens per task than Argon, so the cost per task is higher at the same rate.

Ratings are estimated — limited independent data is available for this model.

Operator guidance

Default to Gemini 4 Argon, which Google says remains the better choice for most software engineering and knowledge work at the same per-token price. Step up to Carbon for the hardest tasks, accepting more tokens (and so more cost) per task. Introductory pricing ($2/$10) runs until 2026-11-30; after that it's $4/$20, so re-check routing decisions in December.

Use cases

Limitations

Citations