Today's price · per 1M tokens
Input
$3.00
per 1M tokens
Output
$15.00
per 1M tokens
Blended
$6.00
blended $/1M — 3:1 weighted input:output
Cached input
$0.150
5% of input — prompt caching
How this price is scoped: Standard paid tier, per 1M tokens: input $3.00, output $15.00 (thinking tokens included), context caching $0.15 (implicit caching only; explicit caching is not supported). Batch and Flex requests cost half: $1.50 input, $7.50 output. Google's Priority tier costs $5.40 input and $27.00 output. We publish the Standard rate.
Price receipt
We read Google's own pricing page on and found Gemini 4 Neon listed with a price on the same card — that read is where the number above comes from. We keep the page text we read; its content hash is 0019bd5780.
Overview
Google's Gemini 4 model for frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work and science workflows, generally available since 2026-11-18. $3/1M input and $15/1M output on the Standard paid tier. 1M-token context shared by input and output, with long-decode output up to 1M tokens and four thinking levels.
Capabilities
struck through = not supportedBenchmark performance
accuracy % · higher is betterSources: Epoch AI (epoch.ai/benchmarks, CC BY 4.0) (GPQA Diamond)
Source: Epoch AI (epoch.ai/benchmarks), CC BY 4.0
- AA Intelligence Index: 61.4 (xhigh effort) (Artificial Analysis composite across reasoning, knowledge and coding evals)
- AA-Omniscience Index: 54.8 (xhigh effort) (−100–100)
- FrontierMath Tiers 1–3: 91.2% (xhigh effort) (Research-level maths (funded by OpenAI, run by Epoch); best run across the effort settings Epoch tested)
- FrontierMath Tier 4: 97.6% (xhigh effort) (Hardest research maths; best run across the effort settings Epoch tested)
- HLE-Diamond: 71.3% (xhigh effort, measured 2026-11-22) — A revised 1,000-question set of Humanity's Last Exam, half reasoning, half knowledge
- CritPt v2, hard subset: 62% (low effort, measured 2026-11-22) — The hardest research-level physics problems in CritPt v2, solved with a code sandbox
- FrontierCode-AA 2.0: 44.4% (xhigh effort, measured 2026-11-22) — Coding tasks in real open-source repositories, graded on whether a maintainer would merge the change
- Terminal-Bench-Science: 70% (xhigh effort, measured 2026-11-22) — Expert-written science research workflows completed in a terminal
- GDP.pdf: 33.4% (high effort, measured 2026-11-22) — Tasks over long professional documents; share of tasks that pass every criterion
- AutomationBench-AA: 79.3% (xhigh effort, measured 2026-11-22) — Multi-step work across simulated business apps, without breaking the rules set
- MLCR-AA: 31.7% (xhigh effort, measured 2026-11-22) — Reasoning over long, fragmented medical records; pass rate
- Terminal-Bench 4.0: 63.6% (xhigh effort, measured 2026-11-22)
- Cost to run the AA Intelligence Index: $6.45 per task, 22.3 min per task (xhigh effort)
- Measured by AA: 105.9 tokens/sec output, 59.77s to first token (xhigh effort)
- Ranks #2 of 207 comparably-measured models by percentile score, across 3 independent measurements
- Ranks #157 of 207 comparably-tested models by normalized performance per dollar
- Strongest at GPQA Diamond — 96.3%, #5 of 128
105.9 tokens/sec output 59.77s latency to first token (TTFT)
Throughput: Artificial Analysis, measured 2026-11-22 · Source: Artificial Analysis (artificialanalysis.ai)
Source: Epoch AI (epoch.ai/benchmarks, CC BY 4.0) · updated · See full rankings →
Specifications
- Provider
- Context window
- 1M tokens
- Modality
- text, image, audio, video
- Parameters
- Proprietary
- Open source
- No — proprietary
- Released
- Status
- Current
- Last updated
- Tags
Availability verified: — listed on Google's own page