Gemini 4 Hydrogen vs Gemini 3.8 Flash: Benchmarks and Pricing Compared
Google has officially announced Gemini 4 Hydrogen, its latest fast model in the Gemini 4 family. Here is how it compares with the popular Gemini 3.8 Flash.
Google's comparison | |||
| Benchmark |  Gemini 4 Hydrogen | Gemini 3.8 Flash | Difference |
|---|---|---|---|
| Terminal-Bench 4.1 | 57.0% | 18.2% | Hydrogen+38.8 pts |
| Task speed at the default setting | 2.5x faster | Baseline | |
| Fabricated claims | More than 5x fewer | Baseline | |
| Enterprise workflows in partner tests | 2x to 2.8x faster | Baseline | |
| Tokens in partner tests | 12% to 23% fewer | Baseline | |
Artificial Analysis benchmarks | |||
| Intelligence Index at xhigh | 54 | Not offered | |
| Intelligence Index at high | 51 | 31 | Hydrogen+20 |
| AA-Briefcase at xhigh | 1567 Elo | Not offered | |
| AA-Briefcase at high | 1497 Elo | 1203 Elo | Hydrogen+294 Elo |
Cost and efficiency | |||
| Cost per Index task at xhigh | $3.16 | Not offered | |
| Cost per Index task at high | $1.87 | $7.77 | Hydrogen76% cheaper |
| Output tokens at high | 110M | 126M | Hydrogen13% fewer |
| Input tokens at high | 5.34B | 10.2B | Hydrogen48% fewer |
Pricing and specs | |||
| Input now, per 1M tokens | $1.50 | $0.75 | 3.8 Flash50% cheaper |
| Output now, per 1M tokens | $7.50 | $3.75 | 3.8 Flash50% cheaper |
| Cached input now, per 1M tokens | $0.075 | $0.075 | |
| Input from January 1, 2027 | $1.50 | $1.50 | |
| Output from January 1, 2027 | $7.50 | $7.50 | |
| Cached input from January 1, 2027 | $0.075 | $0.15 | Hydrogen50% cheaper |
| Context window | 1M tokens (input + output) | 1M tokens | |
| Max output | 64K tokens, up to 1M with long decode | 64K tokens | |
| Thinking levels | Low to xhigh | Low to high | |
| Free API tier | Yes | Yes | |
| Released | November 12, 2026 | September 2, 2026 | Hydrogen71 days newer |