7.6tokens per second
Faster than reading
First word in 6.4 s · a 300-word answer in about 59 s
How fast is Gemma 4 E2B on the Pixel 6 Pro?
About 7.6 tokens per second, faster than reading. The first word arrives after 6.4 seconds.
The same for the Pixel 6 and Pixel 6a, with the same Google Tensor chip. · 6–12 GB
Gemma 4 E2B on other Android phones
Every model on the Pixel 6 ProGemma 4 E2B on every device
All numbers
| Device | Model | Tokens per second | Typical range | First word | Devices | Answers |
|---|---|---|---|---|---|---|
| Pixel 6 Pro Android · Google Tensor | Gemma 4 E2B | 7.6 | 5.9–11 | 6.4 s | 10 | 40 |
| Galaxy S25 Ultra Android · Snapdragon 8 Elite | Gemma 4 E2B | 22 | 15–31 | 0.5 s | 113 | 632 |
| Galaxy S26 Ultra Android · Snapdragon 8 Elite Gen 5 | Gemma 4 E2B | 18 | 14–32 | 0.8 s | 56 | 261 |
| Pixel 10 Pro XL Android · Google Tensor G5 | Gemma 4 E2B | 16 | 13–18 | 1.0 s | 28 | 105 |
| Galaxy S25 FE Android · Exynos 2400 | Gemma 4 E2B | 16 | 14–21 | 1.1 s | 29 | 102 |
| Galaxy S24 Ultra Android · Snapdragon 8 Gen 3 | Gemma 4 E2B | 15 | 13–21 | 1.3 s | 100 | 576 |
| Galaxy S23 Ultra Android · Snapdragon 8 Gen 2 | Gemma 4 E2B | 14 | 9.8–15 | 5.9 s | 54 | 310 |
| Galaxy S22 Ultra Android · Snapdragon 8 Gen 1 | Gemma 4 E2B | 12 | 9.3–14 | 1.8 s | 37 | 213 |
| Pixel 9a Android · Google Tensor G4 | Gemma 4 E2B | 11 | 6.5–13 | 1.3 s | 40 | 147 |
| Galaxy S21 FE 5G Android · Snapdragon 888 | Gemma 4 E2B | 11 | 8.9–13 | 9.0 s | 28 | 124 |
| Galaxy S20 FE 5G Android · Snapdragon 865 / 870 | Gemma 4 E2B | 11 | 8.9–13 | 2.9 s | 14 | 48 |
| moto g stylus 5G Android · Snapdragon 6 Gen 1 | Gemma 4 E2B | 11 | 9.1–11 | 1.2 s | 26 | 135 |
| Pixel 8 Android · Google Tensor G3 | Gemma 4 E2B | 10 | 5.0–12 | 1.3 s | 23 | 91 |
| TCL NXTPAPER 70 Pro Android · Dimensity 7300 | Gemma 4 E2B | 10 | 8.2–11 | 2.3 s | 36 | 171 |
| Galaxy Note10+ Android · Snapdragon 855 / 860 | Gemma 4 E2B | 10 | 8.0–11 | 1.6 s | 14 | 64 |
| moto g stylus Android · Snapdragon 6 Gen 3 | Gemma 4 E2B | 10 | 9.0–11 | 2.2 s | 31 | 189 |
| Pixel 7 Android · Google Tensor G2 | Gemma 4 E2B | 9.1 | 8.6–9.6 | 1.9 s | 12 | 71 |
| Galaxy A35 5G Android · Exynos 1380 | Gemma 4 E2B | 7.7 | 6.9–8.9 | 2.1 s | 10 | 84 |
| moto g power 5G Android · Dimensity 7020 | Gemma 4 E2B | 5.9 | 5.6–6.6 | 3.9 s | 17 | 133 |
| moto g power Android · Dimensity 6100+ / 6300 | Gemma 4 E2B | 5.1 | 4.3–6.2 | 5.4 s | 26 | 137 |
Run Gemma 4 E2B on your phone
Loci is free, needs no account, and keeps every conversation on your device.
Last 30 days, through · updated daily
How these speeds are measured
- Every number comes from real chats in Loci, timed on the device that wrote the answer, with Loci's default settings.
- Tokens per second is how fast an answer is written once it starts. On phones, four characters of text count as one token for every model; on a Mac, Loci counts the model's own tokens. Reading pace is about 5.
- First word is how long an answer takes to start after you send a message. Longer conversations take longer.
- Each device counts once. The large number is the median; the typical range covers the middle half of devices.
- Each phone speed comes from at least 10 phones. Fewer Macs run Loci, so each Mac speed says how many Macs it comes from.
- On phones, answers written in Low Power Mode or Battery Saver, in the background, or with web search or other tools are left out.