Best model for multilingual
Models ranked by their multilingual score, 0 to 100, as on the category page. Only ranked models with enough scored items are included, and models are grouped by total parameters.
Read the memory figure as an estimate, not a guarantee. The memory a model needs is estimated from its file: the weights plus about 20% for the context of a single conversation. It is not a measurement, and a long conversation or several at once needs more. A model with no file size on record is not counted as fitting any amount of memory. Speeds, and the GPU memory shown on the speed page, were measured on this benchmark's server, not on yours. That server kept 1 to 8 answers in flight at a time and loaded a context window of 16,384 tokens for most models; 42 of the models ranked ran at a shorter context. Speeds were measured on one laptop and are comparable only between models on the same backend. See the settings and the memory method.
All sizes
Devstral Small 2 24B Instruct (2512) leads here with 100.0 on a scale of 0 to 100; its Overall is 92.1 (95% interval 89.3–94.7).
| Place | Model | Multilingual (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | Devstral Small 2 24B Instruct (2512)code model | 100.020 of 20 items scored | 92.189.3–94.7 | 9–67 | 3.1ROCm 7.1 | 18.2 | 16–35B |
| 1 | Gemma 4 12B IT | 100.020 of 20 items scored | 98.697.0–99.8 | 1–12 | 8.1ROCm 7.1 | 8.9 | 9–16B |
| 1 | gemma-2-9b-it | 100.020 of 20 items scored | 79.975.5–84.2 | 69–131 | 7.6ROCm 7.1 | 6.9 | 9–16B |
| 1 | gemma-3-12b-it | 100.020 of 20 items scored | 92.789.7–95.4 | 8–66 | 6.1ROCm 7.1 | 9.8 | 9–16B |
| 1 | gemma-3-27b-it | 100.020 of 20 items scored | 95.893.7–97.5 | 3–36 | 2.7ROCm 7.1 | 20.9 | 16–35B |
Tiny (under 3B)
Lfm3 VL 2B leads its size class with 91.7 on a scale of 0 to 100; its Overall is 72.2 (95% interval 67.2–77.0).
| Place | Model | Multilingual (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | Lfm3 VL 2B | 91.720 of 20 items scored | 72.267.2–77.0 | 89–158 | 41.9ROCm 7.1 | 1.8 | Tiny (under 3B) |
| 2 | Qwen3 1.7B | 87.520 of 20 items scored | 76.071.3–80.9 | 77–141 | 38.5ROCm 7.1 | 1.7 | Tiny (under 3B) |
| 3 | Lfm3 1.7B | 83.320 of 20 items scored | 70.265.3–75.3 | 96–162 | 42.1ROCm 7.1 | 1.3 | Tiny (under 3B) |
| 4 | Gemma2 2B Instruct | 80.020 of 20 items scored | 68.262.9–73.3 | 112–160 | 27.6ROCm 7.1 | 2.0 | Tiny (under 3B) |
| 5 | Qwen3 VL 2B Instruct | 75.820 of 20 items scored | 77.472.5–82.1 | 68–137 | 26.5ROCm 7.1 | 2.3 | Tiny (under 3B) |
3–4B
Spark-X2.6-4B leads its size class with 98.3 on a scale of 0 to 100; its Overall is 85.0 (95% interval 80.9–89.1).
| Place | Model | Multilingual (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | Spark-X2.6-4B | 98.320 of 20 items scored | 85.080.9–89.1 | 43–106 | 19.7ROCm 7.1 | 3.0 | 3–4B |
| 2 | Ministral 3 3B Instruct 2512 | 92.520 of 20 items scored | 80.976.5–85.2 | 61–124 | 70.9ROCm 7.1 | 3.6 | 3–4B |
| 3 | granite-4.1-3b | 86.720 of 20 items scored | 80.775.9–85.3 | 66–129 | 20.1ROCm 7.1 | 2.5 | 3–4B |
| 4 | Granite4 Micro H | 83.320 of 20 items scored | 76.672.0–81.4 | 72–139 | 13.4ROCm 7.1 | 2.3 | 3–4B |
| 5 | Llama-3.2-3B-Instruct | 81.720 of 20 items scored | 75.771.0–80.4 | 84–150 | 20.7ROCm 7.1 | 2.4 | 3–4B |
4–9B
gemma-3-4b-it leads its size class with 100.0 on a scale of 0 to 100; its Overall is 82.1 (95% interval 77.6–86.5).
| Place | Model | Multilingual (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | gemma-3-4b-it | 100.020 of 20 items scored | 82.177.6–86.5 | 58–123 | 16.9ROCm 7.1 | 4.0 | 4–9B |
| 1 | Gemma4 E2B IT | 100.020 of 20 items scored | 93.891.0–96.1 | 4–61 | 17.2ROCm 7.1 | 5.5 | 4–9B |
| 1 | Gemma4 E4B IT | 100.020 of 20 items scored | 95.593.0–97.7 | 2–51 | 13.2ROCm 7.1 | 7.9 | 4–9B |
| 1 | Qwen3-4B-Instruct-2507 | 100.020 of 20 items scored | 90.387.2–93.3 | 13–73 | 17.1ROCm 7.1 | 3.0 | 4–9B |
| 1 | Qwen3-8B | 100.020 of 20 items scored | 91.588.4–94.5 | 11–72 | 9.3ROCm 7.1 | 6.3 | 4–9B |
9–16B
Gemma 4 12B IT leads its size class with 100.0 on a scale of 0 to 100; its Overall is 98.6 (95% interval 97.0–99.8).
| Place | Model | Multilingual (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | Gemma 4 12B IT | 100.020 of 20 items scored | 98.697.0–99.8 | 1–12 | 8.1ROCm 7.1 | 8.9 | 9–16B |
| 1 | gemma-2-9b-it | 100.020 of 20 items scored | 79.975.5–84.2 | 69–131 | 7.6ROCm 7.1 | 6.9 | 9–16B |
| 1 | gemma-3-12b-it | 100.020 of 20 items scored | 92.789.7–95.4 | 8–66 | 6.1ROCm 7.1 | 9.8 | 9–16B |
| 1 | Translategemma 12B ITfine-tune of Gemma 3translation model | 100.020 of 20 items scored | 89.686.4–92.9 | 13–80 | 7.6ROCm 7.1 | 9.7 | 9–16B |
| 5 | Ministral 3 14B Instruct (2512) | 98.320 of 20 items scored | 92.790.0–95.2 | 9–61 | 5.6ROCm 7.1 | 10.9 | 9–16B |
16–35B
Devstral Small 2 24B Instruct (2512) leads its size class with 100.0 on a scale of 0 to 100; its Overall is 92.1 (95% interval 89.3–94.7).
| Place | Model | Multilingual (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | Devstral Small 2 24B Instruct (2512)code model | 100.020 of 20 items scored | 92.189.3–94.7 | 9–67 | 3.1ROCm 7.1 | 18.2 | 16–35B |
| 1 | gemma-3-27b-it | 100.020 of 20 items scored | 95.893.7–97.5 | 3–36 | 2.7ROCm 7.1 | 20.9 | 16–35B |
| 1 | gemma-4-26B-A4B-it | 100.020 of 20 items scored | 97.895.8–99.3 | 1–22 | 15.7ROCm 7.1 | 21.6 | 16–35B |
| 1 | gemma-4-31B-it | 100.020 of 20 items scored | 99.398.3–100.0 | 1–5 | 3.8ROCm 7.1 | 22.4 | 16–35B |
| 1 | GLM-4.7-Flash | 100.020 of 20 items scored | 88.985.2–92.5 | 17–84 | 14.1ROCm 7.1 | 22.8 | 16–35B |
> 35B
gpt-oss-120b leads its size class with 100.0 on a scale of 0 to 100; its Overall is 98.6 (95% interval 96.8–100.0).
| Place | Model | Multilingual (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | gpt-oss-120b | 100.020 of 20 items scored | 98.696.8–100.0 | 1–14 | 11.0Vulkan | 78.4 | > 35B |
| 1 | Llama 4 Scout 17B-16E Instruct | 100.020 of 20 items scored | 93.690.7–96.1 | 4–65 | 17.0Vulkan | 80.9 | > 35B |
| 1 | Llama-3.3-70B-Instruct | 100.020 of 20 items scored | 94.391.3–96.7 | 4–63 | 1.4Vulkan | 51.0 | > 35B |
| 1 | Qwen3-Coder-Nextcode model | 100.020 of 20 items scored | 96.494.2–98.3 | 2–42 | 18.1Vulkan | 62.1 | > 35B |
| 1 | Sage-A1.5-Q4_K_M | 100.020 of 20 items scored | 94.591.8–96.7 | 5–56 | 22.6ROCm 7.1 | 26.4 | > 35B |
A higher place is not always a clear win. When two models have overlapping rank ranges, this battery cannot tell them apart; see how sure a score is. A dash means not measured or not scored, which is different from 0. Speeds were measured on one laptop and are comparable only between models on the same backend.
More: all the best-model pages, the hardware picker, speed and memory for every model.