Hydrogen lists at $1.50 / $7.50 per 1M input/output tokens; Claude Sonnet 5.5 at $2 / $10. In Google’s own model-card table Hydrogen scores higher on 7 of 10 benchmarks, while Sonnet 5.5 leads on Terminal-Bench 4.1, OSWorld-2.1 and Terminal-Bench-Science. On the independent Artificial Analysis Intelligence Index, Sonnet 5.5 (max) is 55 and Hydrogen is 54 at xhigh (51 at its default high).
Prices are official list prices retrieved 13 November 2026. Artificial Analysis Intelligence Index figures are from Artificial Analysis model pages retrieved the same day (Hydrogen high = 51 is Google’s citation in its announcement). This site has not run these models. Source: Google AI Studio / Gemini API docs, Retrieved 13 November 2026 · Source: Anthropic, Retrieved 13 November 2026 · Source: Google DeepMind, 12 November 2026 · Source: Google Blog, 12 November 2026
| Gemini 4 Hydrogen | Claude Sonnet 5.5 | |
|---|---|---|
| Input / 1M tokens | $1.50 | $2 |
| Output / 1M tokens | $7.50 (thinking included) | $10 |
| Cache hits / 1M | $0.075 (implicit caching only) | $0.10 (0.05× base input) |
| Context | 1,048,576-token unified window; up to 1M output via background long decode | Full 1M window at standard pricing (Anthropic: Claude 4.6 and later) |
| Free API use | Free tier (rate-limited) in AI Studio / Gemini API | Free credits for new users |
| Artificial Analysis Intelligence Index | 54 (xhigh) · 51 (high, per Google) — AA page | 55 (max) — AA page |
The Gemini 4 Hydrogen model card compares it directly with Claude Sonnet 5.5 and GPT-6.1 Terra. These are vendor-reported numbers; Google links its methodology from the card.
| Category | Benchmark | Notes | Gemini 4 Hydrogen | Claude Sonnet 5.5 |
|---|---|---|---|---|
| Agentic coding | SWE-sweep | 13.5% | 3.1% | |
| Agentic coding | Terminal-Bench 4.1 | Cheating-adjusted / 作弊校正 | 57.0% | 63.9% |
| Knowledge work | Agents' Last Exam v1.1 | 71.6% | 60.4% | |
| Knowledge work | AutomationBench-AA | 76.1% | 71.8% | |
| Knowledge work | Vending-Bench 2 | Bank balance / 余额 | $14,549 | $11,165 |
| Computer use | OSWorld-2.1 | Offline subset / 离线子集 | 77.6% | 83.9% |
| Multimodal | MKA | Climb WSR accuracy | 53.4% | 52.0% |
| Multimodal | Chartography | 66.6% | 61.6% | |
| Science and math | Terminal-Bench-Science 0.1 | 51.4% | 59.9% | |
| Reasoning | HLE-Diamond | 58.8% | 53.3% |
Source: Google DeepMind, 12 November 2026 · Source: Google DeepMind, Cited by the model card, November 2026
10M input + 2M output tokens, no caching or discounts: Hydrogen = 10 × $1.50 + 2 × $7.50 = $30. Sonnet 5.5 = 10 × $2 + 2 × $10 = $40. Per-token prices are not perfectly comparable across vendors: Anthropic’s pricing page says Claude 4.7 and later models use a newer tokenizer that produces about 30% more tokens for the same text.
More comparisons vs Claude Opus 5.5 vs GPT-6.1 Sol vs Gemini 3.1 Pro vs 3.8 Flash vs Argon / Carbon All benchmarks Pricing
Have you tried Hydrogen against one of these? Submit your case — real public posts get listed and ranked.
Author-stated time and cost from public posts, not our benchmark. All same-prompt comparisons →
Same task (prompt not shown identical) · High · @gwammyy, 2026-11-13
| Model | Time | Cost / quota |
|---|---|---|
| Claude Sonnet 5.5 | not stated | not stated |
| Gemini 4 Hydrogen | not stated | not stated |
The author posted left/right outputs labelled Sonnet 5.5 (high) and Hydrogen (high), with the short request 'Oil painting of rocket'. No time or cost is stated.
At its default high level, Hydrogen measured 171 output tokens/s and 8.3 minutes per Artificial Analysis Index task; the fastest level of the other model here measured 144 tok/s, so Hydrogen streamed about 1.2× as fast in AA’s runs.
| Model · thinking/effort level | AA Intelligence Index | Output speed (tok/s) | Minutes per AA Index task | Time to first answer token (s) |
|---|---|---|---|---|
| Gemini 4 Hydrogen · xhigh | 54 | 178 | 14.5 | 44.79 |
| Gemini 4 Hydrogen · high | 51 | 171 | 8.3 | 17.37 |
| Gemini 4 Hydrogen · medium | 41 | 174 | 3.8 | 11.81 |
| Gemini 4 Hydrogen · low | 27 | 173 | 1.4 | 5.58 |
| Claude Sonnet 5.5 · max | 55 | 141 | 33.8 | 272.39 |
| Claude Sonnet 5.5 · xhigh | 50 | 142 | 14.3 | 28.45 |
| Claude Sonnet 5.5 · high | 42 | 144 | 6.7 | 10.38 |
| Claude Sonnet 5.5 · medium | 31 | 129 | 4.0 | 4.50 |
| Claude Sonnet 5.5 · low | 26 | 135 | 2.6 | 2.18 |
| Model | Official list price (input / output per 1M) | Our arithmetic per task (50K in + 8K out) | ×1,000 tasks | AA measured cost per Index task |
|---|---|---|---|---|
| Gemini 4 Hydrogen | $1.50 / $7.50 flat | $0.135 | $135 | $3.16 (xhigh) |
| Claude Sonnet 5.5 | $2 / $10 | $0.180 | $180 | $10.02 (max) |
Output tokens Artificial Analysis recorded while running its whole Intelligence Index (more tokens = more cost and time for the same work): Gemini 4 Hydrogen: xhigh 304M, high 110M, medium 54.4M, low 20.3M; Claude Sonnet 5.5: max 335M, xhigh 111M, high 54.7M, medium 29.4M, low 21.3M.
Source: Artificial Analysis: Gemini 4 Hydrogen · Artificial Analysis: Claude Sonnet 5.5 (retrieved 14 November 2026) · Source: Google AI Studio / Gemini API docs, Retrieved 13 November 2026
Plug in your own token counts: Hydrogen cost calculator · all comparisons.