ModelGoogleConfirmed
Story 1 of 13 in the Tuesday 10 November editionGoogle launches Gemini 4 Carbon at $2 and $10 introductory API pricing
Released globally across Google AI Studio, Android Studio, Gemini for business and Google AI Ultra, Gemini 4 Carbon targets multi-hour coding, cyber defence and scientific discovery while matching Gemini 4 Argon token rates until 30 November 2026.
Edited by Ashutosh Shrivastava
11 Nov 2026, 09:20 IST · 3 min read
Google released Gemini 4 Carbon globally on 10 November 2026 to paid-tier API developers in Google AI Studio and Android Studio, Gemini for business customers, and Google AI Ultra subscribers across Google Antigravity, the Gemini app, and AI Mode in Search. Following an initial 28 October 2026 Fairwind preview for vetted cyber defenders, Carbon builds on Gemini 4 Argon to sustain multi-hour software engineering, vulnerability remediation, and research workflows. Google also expanded its six-month free Antigravity programme for verified scientists to include Carbon alongside Argon.
Pricing, benchmarks and alignment
Through 30 November 2026, Gemini 4 Carbon matches Argon’s introductory API rate of $2 per million input tokens and $10 per million output tokens, with cached input discounted by 95% to $0.10 per million tokens; on 1 December 2026, standard pricing moves to $4 input and $20 output. Across partner and agent evaluations, Carbon ranks first on CursorBench 4.0, Perplexity’s WANDR, Zapier’s AutomationBench at less than half the per-task cost of the next external model, and Balyasny Asset Management’s 23-configuration financial suite, while placing second on the Multimodal Knowledge Agent (MKA) benchmark at less than one-sixth the cost of the top entry.
Google DeepMind also tightened reinforcement-learning oversight against reward hacking and evaluation gaming. Compared with Argon, Carbon reduces internal deployment misalignment alerts by 80% and logs zero successful cheating attempts on audited Terminal-Bench 4.0 runs.
Key numbers
- $2 input and $10 output per million API tokens until 30 November 2026 ($4 and $20 thereafter; $0.10 cached input)
- 40 times higher recall on CASP17 low-homologue protein sequence retrieval than standard JackHMMER pipelines
- 0.0% indirect prompt-injection attack success rate on Gray Swan’s agentic red-teaming suite
- 100% vulnerability scan recall and 95% triage precision across Google’s internal Project Fortify codebases
Technical details
- API model ID and decode cap
- gemini-4-carbon with long-decode support up to 1M tokens (Paid and Enterprise tiers only)
- Standard API pricing
- $2 input, $10 output and $0.10 implicit cached input per 1M tokens until 30 November 2026 ($4, $20, $0.20 from 1 December)
- Batch, Flex and Priority tiers
- $1.00 input and $5.00 output on Batch and Flex, or $3.60 and $18.00 on Priority until 30 November 2026
- Search and Maps grounding
- 5,000 free monthly queries shared across Gemini 3 and newer models, then $14 per 1,000 queries (1,500 daily cap)
- Product surfaces
- Google AI Studio, Android Studio, Antigravity, Gemini for business, Google AI Ultra and Search AI Mode
Release details & context
Service tiers, cyber evaluations and caching constraints
Following Google DeepMind’s 28 October 2026 Fairwind preview, the Gemini Developer API pricing table and Gemini 4 Carbon developer guide detail three paid API lanes for gemini-4-carbon (with no free API tier and implicit context caching only, as explicit caching is unsupported): Standard ($2 and $10), Batch and Flex ($1.00 input and $5.00 output with $0.05 cached input), and Priority ($3.60 input and $18.00 output with $0.18 cached input) through 30 November 2026 before each tier doubles on 1 December 2026.
In the Google DeepMind Gemini 4 Carbon report, Carbon also posts 88.2% on SEC-bench Pro, 71.0% on Collinear CWE-bench, 54.8% on ExploitGym and 1768 Elo on WebDev Arena (+89 over Argon), while powering AlphaProof Code’s binary verification of Signal and Antigravity /swarm. Relative to Argon, Carbon halves unsupported factual claims (2 times reduction), cuts reinforcement-learning alerts by 54%, lowers overreach by 26% and reduces under-service refusals by 29%.
Common questions
Does gemini-4-carbon offer a free API tier or explicit context caching?
No, gemini-4-carbon is restricted to Paid and Enterprise API tiers and supports automatic implicit context caching ($0.10 per 1M tokens until 30 November 2026) rather than explicit cache creation.
When do Gemini 4 Carbon’s introductory API rates expire?
Introductory Argon-matched pricing ($2 input and $10 output on Standard, $1.00 and $5.00 on Batch and Flex, and $3.60 and $18.00 on Priority) runs through 30 November 2026 and doubles on 1 December 2026.
Sources
- PrimaryGemini 4 Carbon is now availableblog.google
- OfficialStart building with Gemini 4 Carbonblog.google
- OfficialGemini 4 Carbon: Our next frontier modeldeepmind.google
- OfficialGemini Developer API pricingai.google.dev
- On XSundar Pichai (@sundarpichai) launch announcement on Xx.com
Google release timeline (24 launches tracked)
- Google unveils confidential TPU 8tx and high-memory TPU 8i Max with open-source TPU Sync
Google introduces confidential TPU 8tx and high-memory TPU 8i Max accelerators alongside Agent Compute Engine and the open-source TPU Sync KV-cache layer.
- Google AI Studio adds zero-setup Google Maps Platform toolkit with free daily sandbox calls
Google embeds a Google Maps Platform toolkit into AI Studio with 100 free daily sandbox calls and no credit card requirement.
- Google launches Gemini Decisions API and Gemini Decision 1 at $0.10 per million input tokens
Google launches the Gemini Decisions API and Gemini Decision 1 at $0.10 per million input tokens and zero output charges, leading Google’s calibration suite.
- Google Antigravity adds built-in browser subagent control across parallel tabs
Google Antigravity ships built-in Browser Use in v2.30.0+ so isolated Gemini Flash subagents test live web apps and parallel tabs inside the side panel.