Best model for pushback resistance
Models ranked by their pushback resistance 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
Cogito 8B V1 Preview Llama leads here with 100.0 on a scale of 0 to 100; its Overall is 83.9 (95% interval 79.6–87.8).
| Place | Model | Pushback resistance (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | Cogito 8B V1 Preview Llamafine-tune of Llama 3.1 | 100.012 of 12 items scored | 83.979.6–87.8 | 47–105 | 11.2ROCm 7.1 | 5.9 | 4–9B |
| 1 | Deepseek R1 8B Llama Distill | 100.012 of 12 items scored | 81.776.9–86.1 | 56–122 | 11.9ROCm 7.1 | 5.9 | 4–9B |
| 1 | DeepSeek-Coder-V2-Lite-Instructcode model | 100.012 of 12 items scored | 79.074.3–83.5 | 69–136 | 11.1ROCm 7.1 | 12.4 | 9–16B |
| 1 | DeepSeek-R1-Distill-Qwen-14B | 100.012 of 12 items scored | 85.381.2–89.2 | 38–106 | 7.1ROCm 7.1 | 10.8 | 9–16B |
| 1 | DeepSeek-R1-Distill-Qwen-32B | 100.012 of 12 items scored | 89.186.0–91.9 | 19–85 | 3.1ROCm 7.1 | 23.8 | 16–35B |
Tiny (under 3B)
Lfm3 1.7B leads its size class with 100.0 on a scale of 0 to 100; its Overall is 70.2 (95% interval 65.3–75.3).
| Place | Model | Pushback resistance (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | Lfm3 1.7B | 100.012 of 12 items scored | 70.265.3–75.3 | 96–162 | 42.1ROCm 7.1 | 1.3 | Tiny (under 3B) |
| 1 | Lfm3 VL 2B | 100.012 of 12 items scored | 72.267.2–77.0 | 89–158 | 41.9ROCm 7.1 | 1.8 | Tiny (under 3B) |
| 3 | Gemma2 2B Instruct | 83.312 of 12 items scored | 68.262.9–73.3 | 112–160 | 27.6ROCm 7.1 | 2.0 | Tiny (under 3B) |
| 3 | Granite3.3 2B | 83.312 of 12 items scored | 68.663.8–73.8 | 111–159 | 28.6ROCm 7.1 | 1.8 | Tiny (under 3B) |
| 3 | Qwen2 Math 1.5B Instructmath model | 83.312 of 12 items scored | 33.128.0–38.2 | 203–222 | 45.2ROCm 7.1 | 1.2 | Tiny (under 3B) |
3–4B
granite-4.1-3b leads its size class with 100.0 on a scale of 0 to 100; its Overall is 80.7 (95% interval 75.9–85.3).
| Place | Model | Pushback resistance (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | granite-4.1-3b | 100.012 of 12 items scored | 80.775.9–85.3 | 66–129 | 20.1ROCm 7.1 | 2.5 | 3–4B |
| 1 | Medgemma1.5 4B ITfine-tune of Gemma 3medical model | 100.012 of 12 items scored | 71.666.6–76.4 | 91–159 | 20.2ROCm 7.1 | 4.0 | 3–4B |
| 1 | Phi4 Mini Reasoning 3.8B | 100.012 of 12 items scored | 70.965.9–76.0 | 97–157 | 48.8ROCm 7.1 | 3.8 | 3–4B |
| 4 | Granite4 Micro H | 91.712 of 12 items scored | 76.672.0–81.4 | 72–139 | 13.4ROCm 7.1 | 2.3 | 3–4B |
| 4 | Granite4.2 3B | 91.712 of 12 items scored | 78.273.7–83.1 | 72–140 | 24.3ROCm 7.1 | 2.7 | 3–4B |
4–9B
Cogito 8B V1 Preview Llama leads its size class with 100.0 on a scale of 0 to 100; its Overall is 83.9 (95% interval 79.6–87.8).
| Place | Model | Pushback resistance (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | Cogito 8B V1 Preview Llamafine-tune of Llama 3.1 | 100.012 of 12 items scored | 83.979.6–87.8 | 47–105 | 11.2ROCm 7.1 | 5.9 | 4–9B |
| 1 | Deepseek R1 8B Llama Distill | 100.012 of 12 items scored | 81.776.9–86.1 | 56–122 | 11.9ROCm 7.1 | 5.9 | 4–9B |
| 1 | Falcon3 7B Instruct | 100.012 of 12 items scored | 76.271.6–81.3 | 71–150 | 12.5ROCm 7.1 | 5.5 | 4–9B |
| 1 | Lfm2.5 8B A1B | 100.012 of 12 items scored | 75.770.8–80.5 | 80–143 | 43.4ROCm 7.1 | 6.2 | 4–9B |
| 1 | Ministral 3 8B Instruct 2512 | 100.012 of 12 items scored | 88.785.2–92.1 | 19–86 | 11.0ROCm 7.1 | 7.2 | 4–9B |
9–16B
DeepSeek-Coder-V2-Lite-Instruct leads its size class with 100.0 on a scale of 0 to 100; its Overall is 79.0 (95% interval 74.3–83.5).
| Place | Model | Pushback resistance (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | DeepSeek-Coder-V2-Lite-Instructcode model | 100.012 of 12 items scored | 79.074.3–83.5 | 69–136 | 11.1ROCm 7.1 | 12.4 | 9–16B |
| 1 | DeepSeek-R1-Distill-Qwen-14B | 100.012 of 12 items scored | 85.381.2–89.2 | 38–106 | 7.1ROCm 7.1 | 10.8 | 9–16B |
| 1 | Gemma 4 12B IT | 100.012 of 12 items scored | 98.697.0–99.8 | 1–12 | 8.1ROCm 7.1 | 8.9 | 9–16B |
| 1 | Ministral 3 14B Instruct (2512) | 100.012 of 12 items scored | 92.790.0–95.2 | 9–61 | 5.6ROCm 7.1 | 10.9 | 9–16B |
| 1 | Olmo2 13B 1124 Instruct | 100.012 of 12 items scored | 69.664.5–74.5 | 97–165 | 6.8ROCm 7.1 | 10.1 | 9–16B |
16–35B
DeepSeek-R1-Distill-Qwen-32B leads its size class with 100.0 on a scale of 0 to 100; its Overall is 89.1 (95% interval 86.0–91.9).
| Place | Model | Pushback resistance (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | DeepSeek-R1-Distill-Qwen-32B | 100.012 of 12 items scored | 89.186.0–91.9 | 19–85 | 3.1ROCm 7.1 | 23.8 | 16–35B |
| 1 | Devstral Small 2 24B Instruct (2512)code model | 100.012 of 12 items scored | 92.189.3–94.7 | 9–67 | 3.1ROCm 7.1 | 18.2 | 16–35B |
| 1 | gemma-3-27b-it | 100.012 of 12 items scored | 95.893.7–97.5 | 3–36 | 2.7ROCm 7.1 | 20.9 | 16–35B |
| 1 | gemma-4-26B-A4B-it | 100.012 of 12 items scored | 97.895.8–99.3 | 1–22 | 15.7ROCm 7.1 | 21.6 | 16–35B |
| 1 | gemma-4-31B-it | 100.012 of 12 items scored | 99.398.3–100.0 | 1–5 | 3.8ROCm 7.1 | 22.4 | 16–35B |
> 35B
Flint 40B A2B leads its size class with 100.0 on a scale of 0 to 100; its Overall is 89.0 (95% interval 85.1–92.6).
| Place | Model | Pushback resistance (0 to 100) | Overall | Rank range | Decode (tok/s) | Memory needed (GB, est.) | Size class |
|---|---|---|---|---|---|---|---|
| 1 | Flint 40B A2B | 100.012 of 12 items scored | 89.085.1–92.6 | 14–90 | 25.9ROCm 7.1 | 27.6 | > 35B |
| 1 | Llama 4 Scout 17B-16E Instruct | 100.012 of 12 items scored | 93.690.7–96.1 | 4–65 | 17.0Vulkan | 80.9 | > 35B |
| 1 | Llama-3.3-70B-Instruct | 100.012 of 12 items scored | 94.391.3–96.7 | 4–63 | 1.4Vulkan | 51.0 | > 35B |
| 1 | Sage-A1.5-Q4_K_M | 100.012 of 12 items scored | 94.591.8–96.7 | 5–56 | 22.6ROCm 7.1 | 26.4 | > 35B |
| 5 | gpt-oss-120b | 91.712 of 12 items scored | 98.696.8–100.0 | 1–14 | 11.0Vulkan | 78.4 | > 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.