Gemini 4 Neon vs Gemini 4 Hydrogen
If both models get every analytics question right, what would justify paying more for Neon?
01 // VERDICT
The short answer
| At a glance | Gemini 4 Neon | Gemini 4 Hydrogen |
|---|---|---|
| List priceinput / output per 1M tokens | $3 / $15 | $1.5 / $7.5 (better) |
Highlighted values mark the lower rate or stronger measured score in the same documented run. · Gemini pricing snapshot (verified 2026-11-19)
EXECUTIVE SUMMARY
Start with Hydrogen for SQL and chart questions like Plotly’s. Both models answered 43 of 43 correctly. Plotly reported $6.36 for Hydrogen and $17.03 for Neon across the full exam at OpenRouter prices. Neon is worth testing on the questions your analysts still have to fix by hand.
Choose Gemini 4 Hydrogen
Your job resembles a read-only analytics assistant and Hydrogen meets your accuracy target on your own warehouse.
Choose Gemini 4 Neon
A pilot on your harder analytical tasks shows enough additional correct answers or saved analyst time to justify the extra inference cost.
02 // KEY FACTS
Key facts
- Plotly tested each model once in Plotly Studio Embedded through OpenRouter. The agent could inspect datasets, query SQL, draw charts and show tables against the same private warehouse.
- Hydrogen ran on November 12 and Neon on November 18, 2026. On November 19 Plotly corrected the rubric for a data-quality question and raised Neon from 42 to 43 correct without rerunning inference.
- The $17.03 Neon and $6.36 Hydrogen figures cover one whole 43-question run each, at OpenRouter prices on the respective test dates. The reports do not identify the thinking level passed to each model.
- On November 19, 2026, Google’s paid Standard rates per million fresh input, cached input and output tokens were $3, $0.15 and $15 for Neon, versus $1.50, $0.075 and $7.50 for Hydrogen. Both use implicit caching.
- DeepMind identifies both Neon and Hydrogen as adaptations of Carbon. They have separate model IDs and both support high and xhigh thinking.
03 // TIER PRICING
API rates by service tier
USD per 1M tokens from vendor pricing pages. 'Not published' means exactly that — we don't extrapolate.
| Service tier | Gemini 4 Neon | Gemini 4 Hydrogen |
|---|---|---|
| Standarddefault on-demand serving | in $3 · cached $0.15 · out $15 implicit cache only (no write fee) | in $1.5 · cached $0.075 · out $7.5 implicit cache only (no write fee) |
| Batchasync jobs, 50% off | in $1.5 · cached $0.075 · out $7.5 implicit cache only (no write fee) | in $0.75 · cached $0.0375 · out $3.75 implicit cache only (no write fee) |
| Flexslower/cheaper serving, 50% off | in $1.5 · cached $0.075 · out $7.5 implicit cache only (no write fee) | in $0.75 · cached $0.0375 · out $3.75 implicit cache only (no write fee) |
| Long-context band | Gemini 4 Neon | Gemini 4 Hydrogen |
|---|---|---|
| Long-context ratesapplies per request above the model threshold | Flat pricing same rates at any input size (up to context window) | Flat pricing same rates at any input size (up to context window) |
USD per 1M tokens. Every input token is billed in exactly one bucket — uncached, cached read, or cache write; a write replaces the input rate for those tokens (official rule), so write rates are per tier as listed. These Gemini models use implicit caching. Enter cache hits as cached input and set explicit cache writes to zero. Gemini 4 Neon has a fixed published rate for each tier. · Google Gemini pricing (verified 2026-11-19)
Gemini service tiers
04 // COST SIMULATOR
Run this comparison on your own numbers
Preloaded with a workload typical for this comparison. Change anything — results recompute locally.
Estimated monthly ledger
browser-localToken-based workload estimate only; it does not measure quality or cost per successful task.
| Model | Per request | Monthly |
|---|---|---|
| Gemini 4 Neon | $0.0393 | $393.00 |
| Gemini 4 Hydrogencheapest | $0.01965 | $196.50 |
Monthly spread between the cheapest and priciest selected model: $196.50.
Breakdown: Gemini 4 Neon · standard tier · standard context
- Fresh input (uncached): 8,000 tok × $3/M$0.024
- Cached input (reads): 2,000 tok × $0.15/M$0.0003
- Cache writes: 0 tok · rate unavailable$0.00
- Output (incl. reasoning tokens): 1,000 tok × $15/M$0.015
- Per request$0.0393
Rates: official pricing · verified 2026-11-19
- Per-token arithmetic only: it does not model quality differences or cross-vendor output-token-count differences for the same task.
- Reasoning tokens are billed as output tokens at the chosen model's output rate.
- Long-context band is decided per request by total input tokens vs the provider threshold, not by monthly volume.
- Cache accounting uses the official mutually-exclusive buckets: each input token is billed once, as uncached input, cached read, OR cache write — writes replace the input rate for those tokens, never add to it.
- Claude rows bill the 5-minute-TTL cache write; the published 1-hour TTL ($8/MTok standard) is not modeled (no TTL input).
- Rows flagged 'derived' are computed from a stated official rule (e.g. Anthropic batch -50%), not read verbatim from a price table.
- Model and tier combinations without a published rate appear as Unsupported.
05 // BENCHMARK LOG
Benchmark results and conditions
- DataAnalyticsBench Gemini 4 Neon
- 43 of 43 correct
Plotly Studio Embedded through OpenRouter, one run of 43 private analytics questions. Plotly regraded the November 18 run on November 19 after correcting a data-quality rubric.
- DataAnalyticsBench full-run inference cost Gemini 4 Neon
- $17.03 for all 43 questions
Plotly reported inference cost at OpenRouter prices on the run date.
- DataAnalyticsBench Gemini 4 Hydrogen
- 43 of 43 correct
Plotly Studio Embedded through OpenRouter, one run of 43 private analytics questions. Plotly checked all 43 answers against its answer key.
Tested · Gemini 4 Hydrogen Benchmark Results for Data Analysis
- DataAnalyticsBench full-run inference cost Gemini 4 Hydrogen
- $6.36 for all 43 questions
Plotly reported inference cost at OpenRouter prices on the run date.
Tested · Gemini 4 Hydrogen Benchmark Results for Data Analysis
06 // BY WORKLOAD
Which model for which workload
Confidence labels: Measured = same-harness evidence · Price-derived = list-price arithmetic · Inferred = reasoned from docs · Insufficient evidence = no comparable public data.
| Workload | Our pick | Confidence | Why |
|---|---|---|---|
| Agent workflows | Gemini 4 Hydrogen | Measured | On Plotly’s read-only analytics tasks, both models reached the same 43-answer quality target and Hydrogen had the lower reported total inference cost. |
| Budget-sensitive / high volume | Gemini 4 Hydrogen | Price-derived | At the November 19 Standard rates, equal token volumes and cache-hit counts cost half as much on Hydrogen. The observed cost ratio on a sequence of tool calls can differ. |
| Quality-first | Decide by bake-off | Insufficient evidence | The published exam gives both models the maximum score. Include difficult joins, ambiguous metrics and missing data from your warehouse when testing for a quality difference. |
QUESTIONS
Frequently asked questions
Why did an early Plotly result give Neon 42 out of 43?
On November 19, 2026, Plotly corrected a rubric that penalized a valid data-quality finding. It regraded existing answers. The corrected Neon score is 43/43, equal to Hydrogen’s.
Does the $17.03 Neon figure apply to each SQL query?
It covers inference for the full 43-question exam at the OpenRouter rates used on November 18. Hydrogen’s $6.36 covers its full November 12 run. Each agent could issue several SQL calls per question.
Were both models tested at xhigh thinking?
Plotly documents a common harness and one run per model. It gives no model-specific thinking setting for these runs. Google offers high and xhigh on both, so record the setting when reproducing the comparison.
CONTINUE EXPLORING
Related
VERIFICATION LOG
Sources & freshness
- Googleaccessed 2026-11-19Gemini 4 Neon API model documentationhttps://aistudio.google.com/docs/models/gemini-4-neonModel ID, token limits, implicit caching, thinking levels and supported tools.
- Googleaccessed 2026-11-19Gemini 4 Hydrogen API documentation (gemini-4-hydrogen specs, 1M window, thinking levels)https://aistudio.google.com/docs/models/gemini-4-hydrogen · published 2026-11-12Input, thinking and output share a 1,048,576-token window. Synchronous output defaults to 65,536 tokens. Hydrogen supports background long decode and implicit caching.
- Plotlyaccessed 2026-11-19Gemini 4 Neon Benchmark Results for Data Analysishttps://plotly.com/blog/gemini-4-neon-plotly-data-analytics-bench/ · published 2026-11-18Updated November 19 after a grading correction. The comparison includes one full 43-question run per model, priced through OpenRouter.
- Plotlyaccessed 2026-11-19Gemini 4 Hydrogen Benchmark Results for Data Analysishttps://plotly.com/blog/gemini-4-hydrogen-plotly-data-analytics-bench/ · published 2026-11-12Plotly reported Hydrogen's score and cost for the same private analytics exam on November 12.
- Plotlyaccessed 2026-11-19Introducing DataAnalyticsBenchhttps://plotly.com/blog/introducing-dataanalyticsbench/ · published 2026-10-14Describes the SQL, table, chart and dataset tools, the grading rubric and the use of OpenRouter prices on each model's run date.
- Google DeepMindaccessed 2026-11-19Gemini 4 Neon model cardhttps://deepmind.google/models/model-cards/gemini-4-neon/ · published 2026-11-18DeepMind identifies Carbon as the basis for Neon and dates the knowledge cutoff to March 2026. The evaluation table includes CursorBench 4.1 and cheating-adjusted Terminal-Bench 4.1.
- Google DeepMindaccessed 2026-11-19Gemini 4 Hydrogen Model Card (November 2026 evaluation table vs Claude Sonnet 5.5)https://deepmind.google/models/model-cards/gemini-4-hydrogen/ · published 2026-11-12DeepMind identifies Hydrogen as adapted from Carbon. The November 12 evaluation table compares Hydrogen with Claude Sonnet 5.5 and GPT-6.1 Terra.
- Googleaccessed 2026-11-19What's new in Gemini 4 Neonhttps://aistudio.google.com/docs/whats-new-gemini-4-neonThe Gemini 4 migration guide covers request compatibility, thinking defaults and implicit cache eligibility.
- Googleaccessed 2026-11-19Gemini Developer API pricinghttps://aistudio.google.com/docs/pricingThe paid tables distinguish fresh input, implicit cache hits and output including thinking. Argon and Carbon have introductory Standard, Batch and Flex rates through November 30, 2026, with the next prices effective December 1. Google lists Priority separately.
Pricing data last verified 2026-11-19. Vendors ship frequently — re-verify before making spend commitments. See the methodology page for update cadence and stale-data flags.