whatmodel.sh
· find the best AI models you can runFilter by your hardware and budget
Filter by task
New #1s:
orRSS
detecting your OS…
↗ share
$ terminal
Score vs release date
x
show
● open weights ◆ API only ─┘ best score so far · x: release month
| # | model | score↓ | where↓ | RAM↓ | speed↓ | sources |
|---|---|---|---|---|---|---|
| 1 |  GPT-6 Astra | 85±1 | api $20.00/M |      +4 | ||
| 2 |  Claude Opus 5.5 | 85±1 | api $8.00/M |   +1 | ||
| 3 | GPT-6.1 Sol | 84±1 | api $4.00/M | | ||
| 4 | Claude Fable 5.1 | 83±1 | api $20.00/M | +1 | ||
| 5 |  Muse Wonder Full | 82±2 | Muse Code only |  | ||
| 6 | Claude Sonnet 5.5 | 82±1 | api $4.00/M | | ||
| 7 |  Gemini 4 Argon | 81±1 | api $4.00/M |  | ||
| 8 | Claude Opus 5 | 81±1 | api $10.00/M | +11 | ||
| 9 | GPT-6 Sol | 81±1 | api $4.00/M | +1 | ||
| 10 | GPT-5.5 Pro | 80±1 | api $67.50/M | | ||
| 11 | GPT-5.6 Sol | 80±1 | api $8.00/M | +14 | ||
| 12 | Claude Fable 5 | 80±1 | api $20.00/M | +5 | ||
| 13 | Claude Mythos Preview (limited preview) | 80±1† | limited preview | | ||
| 14 | GPT-5.4 Pro | 79±1 | api $67.50/M | | ||
| 15 |  DeepSeek V4.1 Flash 1017 | 79±1 | api $0.26/M | | ||
| 16 | Muse Wonder | 79±1 | api $8.00/M | | ||
| 17 | GPT-5.5 | 78±0 | api $11.25/M | +16 | ||
| 18 |  MiMo-V2.6-Pro | 78±1 | api $0.54/M +local | ~443GB | | |
| 19 |  Hy4 preview (770B-A49B) | 77±1 | api $1.25/M +local | ~248GB | | |
| 20 | GPT-5.6 Terra | 77±1 | api $4.50/M | +5 | ||
| 21 |  Qwen3.8 Max (2.4T-A95B) | 77±1 | api $3.00/M +local | ~380GB |  +2 | |
| 22 | Claude Opus 4.8 | 77±0 | api $10.00/M | +11 | ||
| 23 | GPT-5.4 | 76±1 | api $5.63/M | +14 | ||
| 24 | DeepSeek V4.1 Flash | 76±1 | local | ~512GB | ~12 tok/s | |
| 25 | MiMo-V2.6-Flash (309B-A15B) | 76±1 | api $0.17/M +local | ~136GB | | |
| 26 | Muse Spark 1.3 | 76±1 | api $2.00/M | | ||
| 27 | Gemini 3.8 Flash | 76±1 | api $1.50/M |  +5 | ||
| 28 | Claude Haiku 5.5 | 76±1 | api $0.20/M | | ||
| 29 | GPT-5.2 Pro | 76±1 | api $57.75/M | | ||
| 30 |  Kimi K3 | 76±1 | api $6.00/M +local | ~884GB | +7 | |
| 31 | GPT-6 Luna | 75±1 | api $0.20/M | | ||
| 32 | GPT-5.6 Luna | 75±1 | api $0.45/M | +4 | ||
| 33 |  Grok 4.6 | 75±1 | api $3.00/M | | ||
| 34 | Gemini 3.7 Flash | 75±1 | api $1.50/M | +3 | ||
| 35 | DeepSeek V4 Pro (0813) | 75±1 | api $1.63/M +local | ~1010GB | +2 | |
| 36 |  GLM-5.3 | 75±1 | api $2.15/M +local | ~239GB | | |
| 37 | Qwen3.8 Max (0902) | 75±1 | api $3.00/M | | ||
| 38 | Muse Spark Open | 75±1 | api $2.00/M +local | ~190GB | | |
| 39 | DeepSeek V3.2 Speciale | 74±3† | comparison only | | ||
| 40 | Gemini 3.1 Pro (Preview) | 74±0 | api $4.50/M | +23 | ||
| 41 | Muse Spark | 74±2 | comparison only | | ||
| 42 | DeepSeek V4 Flash Vision (Exp) | 74±2 | local | ~104GB | | |
| 43 | Claude Opus 4.7 | 74±1 | api $10.00/M | +9 | ||
| 44 | Grok 4.5 | 74±1 | api $3.00/M | +1 | ||
| 45 | Qwen3.8-Omni-Flash | 74±1† | api $0.23/M | | ||
| 46 | Claude Sonnet 5 | 74±1 | api $4.00/M | +5 | ||
| 47 | Qwen3.8-Flash-Next | 74±1 | local | ~59GB | ~27 tok/s | |
| 48 | Claude Opus 4.6 | 74±1 | api $10.00/M | +22 | ||
| 49 | Muse Spark 1.1 | 74±1 | api $2.00/M | | ||
| 50 | Grok 4.7 | 73±1 | api $3.00/M | | ||
| 51 | Muse Spark 1.2 | 73±1 | api $2.00/M | | ||
| 52 |  Step 5 Preview | 73±1 | comparison only | | ||
| 53 | GPT-5.3-Codex | 73±1 | api $4.81/M | | ||
| 54 | DeepSeek V4 Flash (0731) | 73±1 | api $0.11/M +local | ~92GB | +2 | |
| 55 | Gemini 3.5 Flash | 73±1 | api $3.38/M | +3 | ||
| 56 | Qwen3.7 Max | 73±1 | api $3.75/M | | ||
| 57 | GLM-5.3-Flash | 73±1 | api $0.24/M +local | ~103GB | +2 | |
| 58 | GPT-5.2 | 73±1 | api $4.81/M | +21 | ||
| 59 | Gemini 3.6 Flash | 73±1 | api $1.50/M | +2 | ||
| 60 | GLM-5.2 | 72±1 | api $2.15/M +local | ~121GB | +6 | |
| 61 | Grok 4.20 Multi-Agent (0309) | 72±2 | api $1.56/M | | ||
| 62 |  Dola Seed 2.0 Pro | 72±2 | comparison only | | ||
| 63 | Hunyuan Hy3 (295B-A21B) | 72±1 | api $0.23/M +local | ~94GB | | |
| 64 | Qwen3.8 27B | 72±1 | api $1.60/M +local | ~18GB | ~6.7 tok/s | +3 |
| 65 |  LongCat 2.0 (1.6T-A48B) | 72±2 | api $1.30/M +local | ~2155GB | | |
| 66 | Kimi K2.6 | 72±1 | api $1.71/M +local | ~317GB | +11 | |
| 67 | GPT-5 Pro | 72±1 | api $41.25/M | | ||
| 68 | Gemini 3 Pro (Preview) | 72±1 | comparison only | +15 | ||
| 69 | GPT-5 | 72±1 | api $3.44/M | +22 | ||
| 70 |  Mistral Large 4 (public preview) | 72±1 | api $2.06/M | | ||
| 71 | Qwen3.6 Max Preview | 71±1 | api $2.92/M | | ||
| 72 | Claude Sonnet 4.6 | 71±1 | api $6.00/M | +18 | ||
| 73 | Gemini 3 Flash (Preview) | 71±1 | api $1.13/M | +8 | ||
| 74 | DeepSeek V4 Pro (Preview, Apr 2026) | 71±1 | local | ~604GB | +6 | |
| 75 | Claude Opus 4.5 | 71±1 | api $10.00/M | +14 | ||
| 76 | Qwen3.7 Plus | 71±1 | api $0.70/M | | ||
| 77 | Kimi K2.7 Code | 71±1 | api $1.71/M +local | ~359GB | | |
| 78 | DeepSeek V4 Flash (Preview, Apr 2026) | 71±1 | local | ~103GB | +4 | |
| 79 |  North Small Translate 1.0 | 71±5† | api price unknown | | ||
| 80 |  Solar Open 2 (250B-A15B) | 71±1† | local | ~527GB | | |
| 81 | Grok 4.20 (0309 reasoning) | 71±1 | api $1.56/M | | ||
| 82 | GPT-5.4 mini | 70±1 | api $1.69/M | +6 | ||
| 83 |  Beam 1.0 501B A23B | 70±1 | api price unknown | | ||
| 84 |  Ling 3.0 Flash (124B-A5B) | 70±2† | api $0.090/M +local | ~71GB | | |
| 85 | GPT-5.1 | 70±1 | api $3.44/M | +8 | ||
| 86 | GLM-5.1 | 70±0 | api $2.15/M +local | ~119GB | +11 | |
| 87 | o3-pro | 70±2 | api $35.00/M | | ||
| 88 | Qwen3.5 397B-A17B | 70±0 | api $1.09/M +local | ~65GB | +15 | |
| 89 | Gemini 2.5 Pro Preview (06-05) | 70±2 | comparison only | | ||
| 90 |  KAT-Coder-V2.5-Dev (35B-A3B) | 70±2† | local | ~21GB | | |
| 91 |  Ornith 1.5 35B-A3B | 69±1† | local | ~21GB | ~80 tok/s | |
| 92 | Grok 4.3 | 69±1 | api $1.56/M | | ||
| 93 |  MiniMax-M3 | 69±1 | api $0.48/M +local | ~131GB | +2 | |
| 94 | Qwen3.6 Plus | 69±1 | api $1.13/M | +1 | ||
| 95 | Kimi K2.5 | 69±1 | api $1.20/M +local | ~160GB | +15 | |
| 96 | MiMo-V2-Pro | 69±1 | comparison only | | ||
| 97 | Qwen3-Max (2026-01-23) | 69±1† | comparison only | | ||
| 98 | GLM-5 | 69±1 | api $1.55/M +local | ~119GB | +11 | |
| 99 | MiMo-V2.5-Pro (1T-A42B) | 69±1 | api $0.65/M +local | ~163GB | | |
| 100 | o3 | 69±1 | api $3.50/M | +11 | ||
| 101 | Muse Glimmer 30B | 69±1 | api $0.64/M +local | ~18GB | ~27 tok/s | |
| 102 |  Nemotron 3 Ultra 550B-A55B | 69±1 | api $1.35/M +local | ~89GB | | |
| 103 |  Inkling Small (266B-A12B) | 69±1 | api $0.64/M +local | ~86GB | | |
| 104 | Qwen3.6 27B | 68±1 | api $1.20/M +local | ~18GB | ~16 tok/s | +4 |
| 105 | Ring-2.6-1T | 68±2† | comparison only | | ||
| 106 | Nemotron 3.5 Super VL 120B A12B | 68±1 | local | ~254GB | | |
| 107 | Nemotron 3.5 Bolt 74B A9B | 68±2† | local | ~157GB | | |
| 108 |  A.X K2 (688B-A33B) | 68±1† | local | ~1454GB | | |
| 109 | MiMo-V2.5 (310B-A15B) | 68±1 | api $0.21/M +local | ~100GB |  | |
| 110 | Inkling (975B-A41B) | 68±1 | api $1.73/M +local | ~156GB | | |
| 111 | Qwen3.5 Plus | 68±2 | api $0.90/M | | ||
| 112 | Qwen3.5 122B-A10B | 68±1 | api $1.10/M +local | ~71GB | +5 | |
| 113 | GPT-5.5 Instant | 68±1 | comparison only | | ||
| 114 | Gemini 3.5 Flash-Lite | 68±1 | api $0.85/M | | ||
| 115 | LongCat-Flash-Thinking (560B-A27B) | 68±1† | comparison only | | ||
| 116 | GPT-5.4 nano | 68±1 | api $0.46/M | +1 | ||
| 117 | Claude Sonnet 4.5 | 68±1 | api $6.00/M | +19 | ||
| 118 | GLM-4.7 | 68±1 | api $1.00/M +local | ~58GB | +5 | |
| 119 | GLM-5V-Turbo | 68±2 | api $1.90/M | | ||
| 120 | Qwen3.5 27B | 68±0 | api $0.84/M +local | ~18GB | ~29 tok/s | +9 |
| 121 | Gemini 2.5 Pro Experimental (03-25) | 68±2 | comparison only | | ||
| 122 | Grok 4 (0709) | 68±1 | comparison only | +1 | ||
| 123 | MiMo-V2-Omni | 67±2 | comparison only | | ||
| 124 | GPT-5.2-Codex | 67±2 | comparison only | | ||
| 125 | DeepSeek V3.2 | 67±1 | api $0.29/M +local | ~110GB | +15 | |
| 126 | Step 3.5 Flash (196B-A11B) | 67±1 | api $0.15/M +local | ~118GB | ~18 tok/s | |
| 127 | Qwen3.6 35B-A3B | 67±1 | api $0.31/M +local | ~21GB | ~63 tok/s | +5 |
| 128 | MiniMax-M2.7 | 67±1 | api $0.53/M +local | ~75GB | +2 | |
| 129 | Qwen3-Coder-Next (80B-A3B) | 67±2 | api $0.60/M +local | ~25GB | ~62 tok/s | |
| 130 | Claude Opus 4 | 67±1 | comparison only | +4 | ||
| 131 | GPT-5 mini | 67±1 | api $0.69/M | +16 | ||
| 132 | Ornith 1.0 35B (35B-A3B) | 67±1 | local | ~23GB | ~103 tok/s | |
| 133 | Gemini 2.5 Pro | 67±1 | api $3.44/M | +16 | ||
| 134 | Claude Opus 4.1 | 67±1 | comparison only | +2 | ||
| 135 | Hy3 preview (295B-A21B) | 67±1 | local | ~629GB | | |
| 136 |  Gemma 4 31B | 66±1 | api $0.19/M +local | ~24GB | ~11 tok/s | +21 |
| 137 | Gemini 2.5 Pro Preview (05-06) | 66±1 | comparison only | +2 | ||
| 138 | Gemini 2.5 Pro Preview (03-25) | 66±2 | comparison only | | ||
| 139 | Gemini 3.1 Flash-Lite | 66±1 | api $0.56/M | | ||
| 140 | Kimi K2 Thinking | 66±1 | api $1.07/M +local | ~160GB | | |
| 141 | Qwen3.7 Flash | 66±1 | api $0.055/M | | ||
| 142 |  K-EXAONE 2.0 (750B-A37B) | 66±1† | local | ~791GB | | |
| 143 |  Laguna M.1 | 66±2 | api price unknown +local | ~135GB | | |
| 144 | DeepSeek V3.1 Terminus | 66±1 | local | ~230GB | | |
| 145 | DeepSeek V3.2 Exp | 66±1 | local | ~726GB | | |
| 146 |  GigaChat 3.5 Ultra Reasoning (432B-A28B) | 66±1† | local | ~461GB | | |
| 147 | Qwen3 235B-A22B Thinking 2507 | 66±1 | api $0.97/M +local | ~76GB | +6 | |
| 148 | Kimi K2 Instruct (0711) | 66±1 | api $1.00/M +local | ~152GB | +3 | |
| 149 | Qwen3.5 35B-A3B | 66±0 | api $0.35/M +local | ~21GB | ~50 tok/s | +13 |
| 150 | Step 3.7 Flash (198B-A11B) | 66±1 | api $0.44/M +local | ~104GB | | |
| 151 | o1 (2024-12-17) | 66±1 | api $26.25/M | +13 | ||
| 152 | o4-mini | 66±1 | api $1.93/M | +12 | ||
| 153 | Grok 4.1 Fast (Reasoning) | 66±1 | api $0.28/M | +4 | ||
| 154 | LongCat-Flash-Chat (560B-A27B) | 66±1 | local | ~333GB | | |
| 155 |  Spark-X2.5 4B | 66±1† | local | ~3.5GB | ~114 tok/s | |
| 156 | DeepSeek V3.1 | 66±1 | api $0.42/M +local | ~110GB | +3 | |
| 157 | Qwen3.6 Flash | 66±1 | api $0.56/M | | ||
| 158 | GLM-4.6 | 65±1 | api $1.00/M +local | ~58GB | +1 | |
| 159 | Claude Sonnet 4 | 65±1 | comparison only | +12 | ||
| 160 | Qwen3-Max (2025-09-23) | 65±1 | api $2.40/M | | ||
| 161 | MiniMax-M2.5 | 65±1 | api $0.53/M +local | ~75GB | +3 | |
| 162 | o3-mini | 65±1 | api $1.93/M | +13 | ||
| 163 | MiMo-V2-Flash (309B-A15B) | 65±1 | local | ~103GB | | |
| 164 | Claude 3.7 Sonnet | 65±1 | comparison only | +16 | ||
| 165 | DeepSeek R1 0528 | 65±1 | api $0.91/M +local | ~104GB | +6 | |
| 166 | MiniMax-M2.1 | 65±1 | local | ~79GB | | |
| 167 | ChatGPT-4o (2025-03-26) | 65±2 | comparison only | | ||
| 168 | Nemotron Cascade 2 30B-A3B | 65±1 | local | ~20GB | ~90 tok/s | |
| 169 | MiMo-V2.6-Distill-Qwen-9B | 65±2† | local | ~7.8GB | | |
| 170 | Laguna XS 2.1 (33B-A3B) | 65±2† | local | ~21GB | | |
| 171 | Qwen3 235B-A22B Instruct 2507 | 65±1 | api $0.20/M +local | ~73GB | ~15 tok/s | +3 |
| 172 | K-EXAONE (236B-A23B) | 65±1 | local | ~152GB | | |
| 173 |  INTELLECT-3 | 64±1 | local | ~48GB | | |
| 174 | GLM-4.5 | 64±1 | api $1.00/M +local | ~58GB | +3 | |
| 175 | Mistral Medium 3.5 (128B) | 64±1 | api $3.00/M +local | ~43GB | | |
| 176 | Qwen3.5 Flash | 64±1 | api $0.17/M | | ||
| 177 | MiniMax-M2 | 64±1 | api $0.53/M +local | ~75GB | | |
| 178 | Mistral Medium 3.1 (2508) | 64±2 | comparison only | | ||
| 179 | Nemotron 3 Super 120B-A12B | 64±1 | api $0.28/M +local | ~73GB | ~20 tok/s | |
| 180 | Command A+ (05-2026) | 64±1 | api price unknown +local | ~71GB |  | |
| 181 | Laguna XS.2 | 64±1 | local | ~21GB | | |
| 182 | Claude Haiku 4.5 | 64±1 | api $2.00/M | +9 | ||
| 183 | Gemma 4 26B-A4B | 64±1 | api $0.14/M +local | ~18GB | ~55 tok/s | +13 |
| 184 | EXAONE 4.5 33B | 64±2† | local | ~24GB | | |
| 185 | Solar Pro 4 | 64±1 | api $0.53/M | | ||
| 186 | Qwen3-VL 235B-A22B Thinking | 64±1 | local | ~106GB | | |
| 187 | Kimi K2 Instruct (0905) | 64±2 | local | ~298GB | | |
| 188 | Qwen3-Coder 480B-A35B Instruct | 63±1 | api $0.67/M +local | ~78GB | | |
| 189 | o1-preview | 63±1 | comparison only | +7 | ||
| 190 | Grok 3 Beta | 63±1 | comparison only | | ||
| 191 | Grok 3 Mini Beta | 63±1 | comparison only | +2 | ||
| 192 |  MiniCPM5 2B | 63±1† | local | ~2.4GB | | |
| 193 | Nemotron-Labs-3 Puzzle 75B-A9B | 63±1† | local | ~35GB | | |
| 194 | DeepSeek R1 (Jan '25) | 63±1 | api $4.00/M +local | ~104GB | +15 | |
| 195 | Qwen3-VL 235B-A22B Instruct | 63±1 | local | ~128GB | | |
| 196 | Qwen3-Next 80B-A3B Instruct | 63±1 | api $0.34/M +local | ~26GB | ~55 tok/s | |
| 197 |  Nova 2 Pro (Preview) | 63±2 | api $3.78/M | | ||
| 198 | Gemini 2.5 Flash Preview (09-2025) | 63±1 | comparison only | | ||
| 199 | Qwen3-Next 80B-A3B Thinking | 63±1 | api $0.41/M +local | ~25GB | +3 | |
| 200 |  Granite 4.2 30B | 63±1 | api $0.28/M +local | ~19GB | | |
| 201 | GPT-4.5 Preview | 63±1 | comparison only | +6 | ||
| 202 | Gemini 2.5 Flash Preview (04-17) | 63±1 | comparison only | | ||
| 203 | GPT-4.1 | 62±1 | api $3.50/M | +17 | ||
| 204 | GLM-4.5-Air | 62±1 | api $0.42/M +local | ~35GB | ~23 tok/s | +4 |
| 205 |  Seed-OSS 36B Instruct | 62±1 | local | ~23GB | | |
| 206 | Gemma 4 12B | 62±1 | local | ~9.2GB | ~29 tok/s | |
| 207 |  Trinity-Large-Thinking | 62±2 | api $0.39/M +local | ~141GB | | |
| 208 | Qwen3.5 9B | 62±0 | api $0.11/M +local | ~6.7GB | ~22 tok/s | +19 |
| 209 | Nemotron 3.5 Lightning 30B-A3B | 62±2 | api $0.087/M +local | ~20GB | ~64 tok/s | |
| 210 |  HyperCLOVA X SEED 32B Think | 62±2† | comparison only | | ||
| 211 | MiniMax-M1 (80k) | 62±1 | api $0.96/M +local | ~959GB | | |
| 212 | GLM-4.6V | 62±2 | api $0.45/M +local | ~50GB | | |
| 213 | Magistral Medium 1.2 (2509) | 62±2† | comparison only | | ||
| 214 | Gemini 2.5 Flash | 62±1 | api $0.85/M | +10 | ||
| 215 | Gemini 2.5 Flash Preview (05-20) | 62±1 | comparison only | +1 | ||
| 216 | Gemini 2.0 Flash-Lite (001) | 62±2 | comparison only | | ||
| 217 | Qwen3-VL 32B Thinking | 62±1 | api $0.28/M +local | ~22GB | | |
| 218 | gpt-oss-120b | 62±1 | api $0.26/M +local | ~64GB | ~56 tok/s | +19 |
| 219 |  Apriel 1.6 15B Thinker | 62±1† | local | ~10GB | | |
| 220 | GPT-5 nano | 62±1 | api $0.14/M | +7 | ||
| 221 | Mistral Large 3 (675B-A41B) | 62±1 | api $0.75/M +local | ~108GB | +1 | |
| 222 | Devstral 2 (2512) | 62±2 | local | ~74GB | | |
| 223 | Gemini 2.5 Flash-Lite Preview (06-17) | 62±2 | comparison only | | ||
| 224 | A.X K1 (519B-A33B) | 62±1† | local | ~1091GB | | |
| 225 | o1-mini | 61±1 | comparison only | +12 | ||
| 226 | GLM-4.5V | 61±2 | api $0.90/M +local | ~69GB | | |
| 227 | Devstral Small 2 24B | 61±2 | local | ~16GB | | |
| 228 | MiniMax-M1 (40k) | 61±1† | comparison only | | ||
| 229 | Llama 3.1 Nemotron Ultra 253B v1 | 61±1 | local | ~82GB | | |
| 230 | Gemini 2.0 Pro Experimental (02-05) | 61±1 | comparison only | | ||
| 231 | Grok 4 Fast | 61±1 | comparison only | | ||
| 232 | DeepSeek V3 0324 | 61±1 | local | ~102GB | +11 | |
| 233 | Claude 3.5 Sonnet (2024-10-22) | 61±1 | comparison only | +14 | ||
| 234 | GPT-4.1 mini | 60±1 | api $0.70/M | +13 | ||
| 235 | Gemini 2.0 Flash (001) | 60±1 | comparison only | +7 | ||
| 236 | Qwen3 30B-A3B Instruct 2507 | 60±1 | api $0.15/M +local | ~19GB | ~100 tok/s | +4 |
| 237 | Qwen3 30B-A3B Thinking 2507 | 60±1 | api $0.75/M +local | ~19GB | +8 | |
| 238 |  ERNIE 4.5 300B-A47B | 60±1† | local | ~97GB | | |
| 239 | Qwen3.5 4B | 60±1 | local | ~3.8GB | +11 | |
| 240 | Mistral Medium 3 (2505) | 60±1 | comparison only | | ||
| 241 |  Mellum2.1 12B-A2.5B Thinking | 60±1† | local | ~8.3GB | | |
| 242 |  Gemma 2 9B IT SimPO | 60±1 | local | ~7.6GB | | |
| 243 | Nemotron-Cascade-14B-Thinking | 60±1† | local | ~10GB | | |
| 244 | Gemma 2 9B IT DPO (Princeton) | 60±1 | local | ~7.6GB | | |
| 245 |  Falcon-H1R 7B | 60±1† | local | ~5.3GB | | |
| 246 | Qwen3-VL 32B Instruct | 60±1 | api $0.28/M +local | ~22GB | | |
| 247 | DeepSeek V3 (Dec '24) | 60±1 | api $0.46/M +local | ~337GB | +10 | |
| 248 | Qwen3 235B-A22B (Apr 2025, hybrid) | 60±0 | api $1.23/M +local | ~73GB | +14 | |
| 249 | Gemini 2.0 Flash-Lite Preview (02-05) | 60±2 | comparison only | | ||
| 250 | Granite 4.2 8B | 60±1 | api $0.11/M +local | ~6.8GB | | |
| 251 | Mistral Small 4 (119B-A6.5B) | 60±2 | api $0.26/M +local | ~37GB | | |
| 252 | QwQ-32B | 60±1 | local | ~21GB | +7 | |
| 253 | Solar Open 100B (102B-A12B) | 60±1† | local | ~41GB | | |
| 254 | Qwen3-VL 30B-A3B Thinking | 59±1 | api $0.75/M +local | ~20GB | | |
| 255 | Trinity-Large-Preview | 59±2 | local | ~155GB | | |
| 256 | OpenReasoning-Nemotron-32B | 59±2† | local | ~21GB | | |
| 257 | EXAONE 4.0 32B | 59±1 | local | ~20GB | | |
| 258 |  Phi-4-reasoning-plus 14B | 59±1 | local | ~10GB | | |
| 259 | GLM-4.7-Flash | 59±1 | api $0.0000/M +local | ~18GB | ~44 tok/s | +1 |
| 260 |  Mercury 2 | 58±2 | comparison only | | ||
| 261 |  DiffusionGemma 26B-A4B | 58±1† | comparison only | | ||
| 262 | gpt-oss-20b | 58±1 | api $0.13/M +local | ~13GB | ~79 tok/s | +16 |
| 263 |  Olmo 3.1 32B Think | 58±1 | local | ~21GB | | |
| 264 | Qwen3 14B | 58±1 | api $0.61/M +local | ~10GB | +7 | |
| 265 | Llama 3.3 Nemotron Super 49B v1 | 58±2 | local | ~22GB | | |
| 266 | DeepSeek R1 Distill Llama 70B | 58±1 | api $0.80/M +local | ~30GB | +1 | |
| 267 | GPT-4o (2024-08-06) | 58±1 | api $4.38/M | +23 | ||
| 268 | Nemotron-Cascade-8B | 58±1† | local | ~6.3GB | | |
| 269 | Command A (03-2025) | 58±1 | api $4.38/M +local | ~37GB | | |
| 270 | Olmo 3 32B Think | 58±1 | local | ~21GB | | |
| 271 | GPT-4o (2024-05-13) | 58±0 | api $7.50/M | +19 | ||
| 272 | Phi-4-reasoning 14B | 58±1 | api $0.22/M +local | ~10GB | | |
| 273 | GPT-4o (2024-11-20) | 58±1 | api $4.38/M | +7 | ||
| 274 | Claude 3.5 Sonnet (2024-06-20) | 58±1 | comparison only | +16 | ||
| 275 | Nova 2 Lite | 58±1 | api $0.94/M | | ||
| 276 | Magistral Small 1.2 | 58±1 | api $0.75/M +local | ~16GB | | |
| 277 | Qwen3-VL 30B-A3B Instruct | 58±1 | api $0.35/M +local | ~20GB | | |
| 278 | Qwen3-VL 8B Thinking | 58±1 | api $0.66/M +local | ~6.9GB | | |
| 279 | Qwen3 32B | 58±0 | api $0.13/M +local | ~20GB | +17 | |
| 280 | Gemini 2.0 Flash Thinking Experimental (01-21) | 58±1 | comparison only | | ||
| 281 | Qwen2.5-Max (2025-01-25) | 58±1 | comparison only | | ||
| 282 | Qwen3 Coder 30B-A3B Instruct | 58±1 | api $0.12/M +local | ~19GB | ~93 tok/s | |
| 283 | Qwen2.5 Coder 32B Instruct | 57±1 | api $0.74/M +local | ~21GB | +5 | |
| 284 | Gemini 2.5 Flash-Lite | 57±2 | api $0.17/M | +1 | ||
| 285 | Llama 4 Maverick (402B-A17B) | 57±1 | api $0.42/M +local | ~65GB | +14 | |
| 286 | DeepSeek R1 0528 Qwen3 8B | 57±1 | api $0.068/M +local | ~6.5GB | ~21 tok/s | |
| 287 | Gemini 1.5 Pro Experimental (0827) | 57±1 | comparison only | | ||
| 288 | Granite 4.2 3B | 57±1 | api $0.052/M +local | ~3.3GB | | |
| 289 | Gemini 1.5 Pro (002) | 57±1 | comparison only | +5 | ||
| 290 | Llama 3.3 Nemotron Super 49B v1.5 | 57±1 | local | ~18GB | | |
| 291 | Nemotron 3 Nano 30B-A3B | 57±1 | api $0.087/M +local | ~20GB | ~90 tok/s | +3 |
| 292 |  Athene-V2-Chat 72B | 57±1 | local | ~44GB | | |
| 293 | Granite 4.1 30B | 57±1† | local | ~19GB | ~25 tok/s | |
| 294 | Grok 2 (1212) | 57±1 | comparison only | | ||
| 295 | Nemotron-Cascade-8B-Thinking | 57±2 | local | ~6.3GB | | |
| 296 | Gemini 2.0 Flash Experimental | 57±1 | comparison only | | ||
| 297 | Qwen2.5-Omni 7B | 57±2 | local | ~5.8GB | | |
| 298 |  Sky-T1 32B Preview | 57±4 | local | ~21GB | | |
| 299 | Ministral 3 14B Reasoning 2512 | 57±1 | local | ~10GB | | |
| 300 | Qwen Plus (2025-01-25) | 57±1 | api $0.60/M | | ||
| 301 | Mistral Small 3.2 24B | 57±1 | api $0.11/M +local | ~15GB | +4 | |
| 302 | DeepSeek-V2.5 (1210) | 57±1 | local | ~82GB | | |
| 303 |  QED-Nano | 57±2 | local | ~4.1GB | | |
| 304 | Qwen3 4B Thinking 2507 | 56±1 | local | ~4.2GB | ~67 tok/s | +3 |
| 305 | GPT-4 Turbo (2024-04-09) | 56±0 | api $15.00/M | +18 | ||
| 306 |  Yi-Lightning | 56±1 | api price unknown | | ||
| 307 | Gemini 1.5 Pro Experimental (0801) | 56±1 | comparison only | | ||
| 308 | Qwen3-VL 4B Thinking | 56±1 | local | ~4.4GB | | |
| 309 | Qwen2.5 72B Instruct | 56±1 | api $0.37/M +local | ~31GB | +24 | |
| 310 | Qwen2.5-VL 72B Instruct | 56±1 | local | ~46GB | +3 | |
| 311 | Llama 3.1 405B Instruct | 56±1 | api $2.40/M +local | ~160GB | +13 | |
| 312 | Mistral Large 2 (2411) | 56±1 | local | ~77GB | +8 | |
| 313 | Qwen2.5 Coder 14B Instruct | 56±1 | local | ~11GB | ~18 tok/s | |
| 314 | DeepSeek R1 Distill Qwen 32B | 56±1 | api $1.59/M +local | ~21GB | ~7.3 tok/s | +5 |
| 315 | Qwen2.5 32B Instruct | 56±1 | local | ~21GB | +11 | |
| 316 | Qwen3 8B | 56±1 | api $0.31/M +local | ~6.3GB | ~25 tok/s | +15 |
| 317 | DeepSeek R1 Distill Qwen 14B | 56±1 | local | ~11GB | ~15 tok/s | +2 |
| 318 | Nemotron Nano 9B v2 | 56±1 | api $0.10/M +local | ~7.6GB | | |
| 319 | Magistral Medium 1.0 (2506) | 56±1 | comparison only | | ||
| 320 | Molmo 2 8B | 56±2 | local | ~6.8GB | | |
| 321 | Ministral 3 14B | 56±1 | api $0.20/M +local | ~9.1GB | ~46 tok/s | +2 |
| 322 | EXAONE Deep 32B | 56±2† | local | ~22GB | | |
| 323 | Qwen3 4B Instruct 2507 | 55±1 | local | ~4.2GB | ~107 tok/s | +4 |
| 324 | Qwen3 30B-A3B (Apr 2025, hybrid) | 55±0 | api $0.35/M +local | ~18GB | +10 | |
| 325 |  Bielik 11B v3.0 Instruct | 55±2 | local | ~8.1GB | | |
| 326 | Nova Premier | 55±2† | api $5.00/M |  | ||
| 327 | Granite 4.1 8B | 55±1 | api $0.11/M +local | ~7.1GB | ~70 tok/s | |
| 328 | Magistral Small 1.0 (2506) | 55±1 | local | ~16GB | | |
| 329 | Spark-X2.5 1.7B | 55±1† | local | ~1.9GB | ~166 tok/s | |
| 330 | Llama 3.3 70B Instruct | 55±1 | api $0.72/M +local | ~25GB | ~13 tok/s | +21 |
| 331 | Magistral Medium 1.1 (2507) | 55±1† | comparison only | | ||
| 332 | Qwen3-VL 8B Instruct | 55±1 | api $0.31/M +local | ~6.9GB | | |
| 333 | Qwen2.5-Math 72B Instruct | 55±2† | local | ~44GB | | |
| 334 | Mistral Large 2 (2407) | 55±1 | local | ~45GB | +10 | |
| 335 | Falcon-H1 34B Instruct | 55±3 | local | ~23GB | | |
| 336 | Llama 4 Scout (109B-A17B) | 55±1 | api $0.15/M +local | ~36GB | ~19 tok/s | +8 |
| 337 | GPT-4 Turbo Preview (1106) | 55±1 | comparison only | +5 | ||
| 338 | Llama 3.1 Tulu 3 70B | 55±1 | local | ~43GB | | |
| 339 | Olmo 3.1 32B Instruct | 55±2 | local | ~21GB | | |
| 340 | Bielik 11B v2.6 Instruct | 55±3 | local | ~8.1GB | | |
| 341 | GigaChat 3.5 Ultra Instruct (432B-A28B) | 55±1† | local | ~457GB | | |
| 342 | Mistral Small 3.1 24B (2503) | 55±1 | local | ~16GB | +7 | |
| 343 | Pixtral Large (2411) | 55±3 | local | ~51GB | | |
| 344 | GPT-4 Turbo Preview (0125) | 55±1 | comparison only | +3 | ||
| 345 | DeepSeek-V2.5 | 55±1 | local | ~151GB | +1 | |
| 346 |  OpenThinker2 32B | 55±1 | local | ~21GB | | |
| 347 | Claude 3.5 Haiku | 55±1 | comparison only | +9 | ||
| 348 | DeepSeek-V2 Chat (0628) | 55±1 | local | ~66GB | | |
| 349 | OpenThinker 32B | 55±1† | local | ~21GB | | |
| 350 | Command A Reasoning (08-2025) | 55±2 | local | ~34GB | | |
| 351 | Magistral Small 1.1 (2507) | 55±1† | local | ~16GB | | |
| 352 | Mellum2 12B-A2.5B Thinking | 54±1† | local | ~8.6GB | | |
| 353 | Ministral 3 8B Reasoning 2512 | 54±1 | local | ~6.5GB | | |
| 354 | Llama 3.1 Tulu 3 70B DPO | 54±1 | local | ~22GB | | |
| 355 | Gemini 1.5 Flash Experimental (0827) | 54±1 | comparison only | | ||
| 356 | Nemotron 3 Nano 4B | 54±1 | local | ~2.9GB | ~52 tok/s | |
| 357 | GPT-4o mini (2024-07-18) | 54±0 | api $0.26/M | +32 | ||
| 358 | Gemini 1.5 Pro (001) | 54±1 | comparison only | +3 | ||
| 359 | DeepSeek-Coder-V2 Instruct (0614) | 54±1 | local | ~92GB | +2 | |
| 360 | DeepSeek-Coder-V2 Instruct (0724) | 54±1 | local | ~151GB | | |
| 361 | Mistral Small 3 24B (2501) | 54±1 | local | ~15GB | +6 | |
| 362 | GPT-4 (0314) | 54±1 | comparison only |  +21 | ||
| 363 | Gemini 1.5 Flash (002) | 54±1 | comparison only | +5 | ||
| 364 | Qwen Turbo (2024-11-01) | 54±2 | comparison only | | ||
| 365 |  GLM-4 9B Chat 1M | 54±4 | local | ~8.1GB | | |
| 366 | EXAONE Deep 7.8B | 54±2† | local | ~6.5GB | | |
| 367 | Claude 3 Opus | 54±0 | comparison only | +15 | ||
| 368 | Qwen2-VL 72B Instruct | 54±1 | local | ~34GB | | |
| 369 |  Hermes 3 Llama 3.1 405B | 54±1 | local | ~114GB | | |
| 370 | Yi-Large Preview | 54±1 | comparison only | | ||
| 371 | Llama 3.1 70B Instruct | 54±0 | api $0.72/M +local | ~27GB | +22 | |
| 372 | Gemma 3 27B | 54±1 | api $0.10/M +local | ~19GB | +23 | |
| 373 | GPT-4.1 nano | 53±1 | api $0.17/M | +11 | ||
| 374 | Qwen2.5-VL 32B Instruct | 53±2 | local | ~23GB | +1 | |
| 375 | Phi-4 14B | 53±1 | api $0.087/M +local | ~10GB | +11 | |
| 376 | GLM-4 (0520) | 53±1 | comparison only | | ||
| 377 | Bielik 11B v2.1 Instruct | 53±3 | local | ~8.5GB | | |
| 378 | Qwen2.5-Math 7B Instruct | 53±2† | local | ~5.4GB | | |
| 379 | Olmo 3 7B Think | 53±1 | local | ~5.8GB | | |
| 380 |  s1.1 32B | 53±1 | local | ~22GB | | |
| 381 | Gemma 4 E4B | 53±1 | local | ~6.4GB | ~21 tok/s | +8 |
| 382 | Ministral 3 8B | 53±1 | api $0.15/M +local | ~6.5GB | ~68 tok/s | +3 |
| 383 | Llama 3.1 Nemotron 70B Instruct | 53±1 | local | ~43GB | +4 | |
| 384 | Qwen3-VL 4B Instruct | 53±1 | local | ~4.4GB | | |
| 385 | GPT-4 (0613) | 53±1 | api $37.50/M | +11 | ||
| 386 |  LFM2.5 2.6B | 53±1 | local | ~2.3GB | | |
| 387 | Bielik 11B v2.2 Instruct | 53±3 | local | ~8.5GB | | |
| 388 | Llama 3.1 Nemotron 51B Instruct | 53±2 | local | ~34GB | | |
| 389 | Nemotron-4 340B Instruct | 53±2 | comparison only | | ||
| 390 | ERNIE 4.5 21B-A3B | 53±1 | comparison only | | ||
| 391 |  Palmyra X4 (palmyra-x-004) | 53±3 | comparison only | | ||
| 392 | Granite 4.0 H Small (32B-A9B) | 53±1 | api $0.11/M +local | ~21GB | +1 | |
| 393 | Qwen2.5 Coder 7B Instruct | 53±1 | local | ~5.4GB | ~3.8 tok/s | |
| 394 | Aya Expanse 32B | 53±2 | api $0.75/M +local | ~22GB | +1 | |
| 395 | LFM2.5 8B-A1B | 53±1 | local | ~5.5GB | ~174 tok/s | |
| 396 |  OpenCoder-8B-Instruct | 53±1 | local | ~6.4GB | | |
| 397 | Gemma 3 12B | 52±1 | api $0.075/M +local | ~10GB | ~13 tok/s | +10 |
| 398 | Qwen2.5 14B Instruct | 52±1 | local | ~11GB | +7 | |
| 399 | Yi-Large | 52±1 | comparison only | | ||
| 400 | Llama-3-Instruct-8B-SimPO-ExPO | 52±1 | local | ~6.6GB | | |
| 401 |  Cogito v1 Preview Llama 8B | 52±4 | local | ~6.2GB | | |
| 402 | Llama-3-Instruct-8B-SimPO | 52±1 | local | ~6.6GB | | |
| 403 | Athene-70B | 52±1 | local | ~29GB | | |
| 404 |  Jamba 1.5 Large | 52±1 | local | ~839GB | | |
| 405 | Qwen2 72B Instruct | 52±1 | local | ~33GB | +14 | |
| 406 |  CodeQwen1.5 7B Chat | 52±1 | local | ~6.4GB | | |
| 407 | Command A Translate (08-2025) | 52±4† | api price unknown | | ||
| 408 |  OpenCodeInterpreter-DS-33B | 52±1 | local | ~22GB | | |
| 409 | Gemini 1.5 Flash (001) | 51±1 | comparison only | +3 | ||
| 410 | Bielik 11B v2.0 Instruct | 51±3 | local | ~8.5GB | | |
| 411 | Flan-U-PaLM 540B | 51±1† | never released | | ||
| 412 | LFM2 24B-A2B | 51±1 | local | ~14GB | | |
| 413 | Qwen3 4B | 51±1 | local | ~4GB | +8 | |
| 414 | DeepSeek-Coder 33B Instruct | 51±1 | local | ~21GB | +1 | |
| 415 | Codestral 22B v0.1 | 51±1 | local | ~15GB | | |
| 416 | GLM-4 (0116) | 51±1 | comparison only | | ||
| 417 | Gemma 2 27B IT | 51±1 | local | ~19GB | +10 | |
| 418 | Llama 3 70B Instruct | 51±0 | local | ~24GB | +17 | |
| 419 | Llama 3.2 90B Vision Instruct | 51±1 | local | ~58GB | +1 | |
| 420 |  WizardCoder 33B V1.1 | 51±1 | local | ~22GB | | |
| 421 | Bielik 11B v2.3 Instruct | 51±2 | local | ~8.6GB | | |
| 422 | Llama-3-8B-ProLong-512k-Instruct | 51±4 | local | ~35GB | | |
| 423 | Flan-PaLM 540B | 51±1† | never released | | ||
| 424 | Nova Pro | 51±1 | api $1.40/M | | ||
| 425 | Qwen2-Math 72B Instruct | 51±1 | local | ~32GB | | |
| 426 | WizardLM-2 8x22B | 51±1 | local | ~91GB | | |
| 427 | Mistral Small Instruct 22B (2409) | 51±2 | local | ~15GB | | |
| 428 | OpenThinker3 7B | 51±1 | local | ~6GB | | |
| 429 | Granite 4.1 3B | 51±1† | local | ~3.7GB | ~131 tok/s | |
| 430 |  Artigenz-Coder-DS-6.7B | 51±1 | local | ~5.8GB | | |
| 431 | OpenCodeInterpreter-DS-6.7B | 51±1 | local | ~5.8GB | | |
| 432 | Mellum2 12B-A2.5B Instruct | 51±2† | local | ~7.6GB | | |
| 433 | Molmo 72B (0924) | 51±2 | local | ~309GB | | |
| 434 | Yi 34B 200K (base) | 51±4 | local | ~23GB | | |
| 435 | Hermes 3 Llama 3.1 70B | 51±1 | local | ~43GB | | |
| 436 | OLMo 2 32B Instruct (0325) | 50±1 | local | ~21GB | | |
| 437 |  speechless-codellama-34b-v2.0 | 50±1 | local | ~23GB | | |
| 438 | Ministral 3 3B Reasoning 2512 | 50±1 | local | ~3.4GB | | |
| 439 |  Aya 23 35B | 50±2 | local | ~22GB | | |
| 440 | AceReason-Nemotron-1.0-14B | 50±2† | local | ~10GB | | |
| 441 | Claude 3 Sonnet | 50±0 | comparison only | +12 | ||
| 442 | Qwen2.5-VL 7B Instruct | 50±1 | local | ~5.8GB | | |
| 443 | Mixtral 8x22B Instruct v0.1 | 50±1 | local | ~56GB | +9 | |
| 444 |  Magicoder-S-DS 6.7B | 50±1 | local | ~5.8GB | | |
| 445 | EXAONE Deep 2.4B | 50±2† | local | ~3.2GB | | |
| 446 | Pixtral 12B (2409) | 50±2 | local | ~9.6GB | | |
| 447 | Nova Lite | 50±2 | api $0.10/M | | ||
| 448 |  XwinCoder-34B | 50±1 | local | ~23GB | | |
| 449 | Qwen2.5 7B Instruct | 50±1 | api $0.30/M +local | ~5.4GB | ~54 tok/s | +12 |
| 450 | LFM2.5 1.2B Thinking | 50±2 | local | ~1.3GB | | |
| 451 | Yi-Coder 9B Chat | 50±1 | local | ~6.3GB | | |
| 452 |  InternLM2.5 7B Chat | 50±2 | local | ~6.1GB | | |
| 453 |  StarChat2 15B v0.1 | 50±1 | local | ~12GB | | |
| 454 |  Code Millenials 34B | 50±1 | local | ~23GB | | |
| 455 | Qwen2-Math 7B Instruct | 50±2† | local | ~6GB | | |
| 456 | WaveCoder-Ultra-6.7B | 50±1 | local | ~5.8GB | | |
| 457 |  Reka Core (2024-05-01) | 50±1 | comparison only | | ||
| 458 | GPT-4o Audio Preview (2024-12-17) | 50±2 | comparison only | | ||
| 459 |  MegaBeam-Mistral-7B-512k | 49±4 | local | ~5.9GB | | |
| 460 | Gemini 1.5 Flash-8B (001) | 49±2 | comparison only | | ||
| 461 | Phi-3.5-MoE-instruct | 49±1 | local | ~25GB | | |
| 462 | Qwen3-VL 2B Thinking | 49±1 | local | ~2.7GB | | |
| 463 | Gemma 3n E4B | 49±1 | local | ~5.6GB | | |
| 464 | Claude 3 Haiku | 49±0 | comparison only | +16 | ||
| 465 | DeepSeek-Coder 7B Instruct v1.5 | 49±1 | local | ~5.9GB | | |
| 466 | OpenThinker2 7B | 49±2† | local | ~6.4GB | | |
| 467 | DeepSeek-Coder 6.7B Instruct | 49±1 | local | ~8.8GB | | |
| 468 | Hermes 2 Theta Llama-3 70B | 49±2 | local | ~47GB | | |
| 469 | Qwen2 57B-A14B Instruct | 49±2 | local | ~35GB | | |
| 470 | MiniCPM3-4B | 49±2 | local | ~4GB | | |
| 471 | Granite 4.0 H Tiny (7B-A1B) | 49±1 | local | ~4.5GB | ~12 tok/s | |
| 472 | Claude 1.3 | 49±1 | comparison only | +1 | ||
| 473 | Yi-1.5 34B (base) | 49±2 | local | ~22GB | | |
| 474 | Bielik 4.5B v3.0 Instruct | 49±2 | local | ~4.3GB | | |
| 475 | Qwen2.5 Coder 3B Instruct | 49±1 | local | ~2.6GB | ~9.5 tok/s | |
| 476 | Gemma 2 9B IT | 49±1 | local | ~8.9GB | +14 | |
| 477 |  WhiteRabbitNeo-33B-v1 | 49±1 | local | ~22GB | | |
| 478 | GPT-4o mini Audio Preview (2024-12-17) | 49±2 | comparison only | | ||
| 479 | Qwen1.5 110B Chat | 49±1 | local | ~45GB | +1 | |
| 480 | Command R Plus (08-2024) | 49±1 | api $4.38/M +local | ~63GB | +1 | |
| 481 | GLM-4 9B Chat | 49±2 | local | ~7.1GB | +1 | |
| 482 | Magicoder-S-CL-7B | 48±1 | local | ~5.8GB | | |
| 483 | speechless-coder-ds-6.7b | 48±1 | local | ~5.8GB | | |
| 484 |  Phind CodeLlama 34B v2 | 48±1 | local | ~23GB | | |
| 485 | AceReason-Nemotron-1.1-7B | 48±2† | local | ~6GB | | |
| 486 | Jamba 1.5 Mini | 48±1 | local | ~34GB | | |
| 487 | DeepSeek R1 Distill Qwen 7B | 48±1 | local | ~5.4GB | ~35 tok/s | +7 |
| 488 | U-PaLM 540B | 48±1† | never released | | ||
| 489 | GPT-3.5 Turbo (0125) | 48±1 | api $0.75/M | +11 | ||
| 490 | Mistral Large (2402) | 48±1 | comparison only | +3 | ||
| 491 |  InternVL2-26B | 48±2 | local | ~55GB | | |
| 492 | Aya Expanse 8B | 48±2 | local | ~6.4GB | | |
| 493 |  EuroLLM 9B Instruct | 48±2 | local | ~6.9GB | | |
| 494 | Flan-UL2 20B | 48±4 | comparison only | | ||
| 495 | GPT-3.5 Turbo (1106) | 48±1 | comparison only | +2 | ||
| 496 | Code Llama 70B Instruct | 48±1 | local | ~28GB | +1 | |
| 497 | Command R (08-2024) | 48±1 | api $0.26/M +local | ~21GB | +3 | |
| 498 | Falcon-H1 7B Instruct | 48±3 | local | ~6GB | | |
| 499 | Granite 4.0 H Micro 3B | 48±1 | local | ~3.4GB | | |
| 500 | Minerva 540B | 48±1† | never released | | ||
| 501 | OLMo 2 13B Instruct (1124) | 48±1 | local | ~9.6GB | | |
| 502 | Llama 3 70B (base) | 48±2 | local | ~43GB | | |
| 503 | InternLM2.5 7B Chat 1M | 48±4 | local | ~6.4GB | | |
| 504 | Qwen3.5 2B | 47±1 | local | ~2.2GB | ~137 tok/s | +7 |
| 505 | GPT-3.5 Turbo (0301) | 47±1 | comparison only | +24 | ||
| 506 | CogVLM2 Llama3 Chinese Chat 19B | 47±7† | comparison only | | ||
| 507 | Hermes 2 Pro Llama 3 8B | 47±3 | local | ~6.6GB | | |
| 508 | MedGemma 4B IT | 47±2 | local | ~4.1GB | | |
| 509 | Nova Micro | 47±2 | api $0.061/M | | ||
| 510 | Yi 1.5 34B Chat | 47±1 | local | ~22GB | +5 | |
| 511 |  OpenChat 3.5 0106 | 47±1 | local | ~6.1GB | +1 | |
| 512 | WizardCoder Python 34B V1.0 | 47±1 | local | ~22GB | | |
| 513 | Tulu 2 70B | 47±2† | local | ~28GB | | |
| 514 | Codex code-davinci-002 (GPT-3.5 base) | 47±0 | comparison only | +20 | ||
| 515 | Gemma 4 E2B | 47±1 | api $0.050/M +local | ~4.5GB | +6 | |
| 516 | Claude 2.0 | 47±1 | comparison only | +2 | ||
| 517 | Gemini 1.0 Pro (001) | 47±2 | comparison only | | ||
| 518 |  DBRX Instruct | 47±1 | local | ~85GB | +3 | |
| 519 | Mistral Medium (2312) | 47±1 | comparison only | | ||
| 520 | GPT-3.5 Turbo (0613) | 47±1 | comparison only | +5 | ||
| 521 | Granite 34B Code Instruct (8K) | 47±1 | local | ~22GB | | |
| 522 | Granite 4.0 Micro 3B | 47±1 | local | ~3.5GB | ~6.2 tok/s | |
| 523 | EXAONE 4.0 1.2B | 47±1† | local | ~2.2GB | | |
| 524 | InternLM2.5 20B Chat | 47±2 | local | ~14GB | | |
| 525 | Cogito v1 Preview Llama 3B | 47±4 | local | ~3.4GB | | |
| 526 | Qwen2.5-Math 1.5B Instruct | 47±2† | local | ~1.8GB | | |
| 527 | Flan-cont-PaLM 62B | 47±1† | never released | | ||
| 528 | Qwen1.5 72B Chat | 47±1 | local | ~31GB | +9 | |
| 529 |  StarCoder2-15B-Instruct-v0.1 | 47±1 | local | ~11GB | | |
| 530 | speechless-starcoder2-15b | 47±1 | local | ~12GB | | |
| 531 | OpenChat 3.5 0106 Gemma | 47±2 | local | ~6.9GB | | |
| 532 | Granite 3.3 8B Instruct | 47±1 | local | ~6.3GB | | |
| 533 | Phi-3 Medium 128K Instruct | 47±1 | local | ~10GB | | |
| 534 | Qwen2-VL 7B Instruct | 47±1 | local | ~5.8GB | | |
| 535 | Zephyr ORPO 141B-A35B v0.1 | 47±2 | local | ~56GB | | |
| 536 | LFM2 8B-A1B | 47±1 | comparison only | | ||
| 537 | PaLM 540B | 47±0† | never released | +5 | ||
| 538 | DeepSeek R1 Distill Llama 8B | 47±1 | local | ~6.2GB | +2 | |
| 539 | Claude 2.1 | 47±1 | comparison only | | ||
| 540 | Phi-3-medium-4k-instruct | 47±1 | local | ~10GB | +1 | |
| 541 | QwQ-32B-Preview | 46±2 | local | ~21GB | +2 | |
| 542 | Mistral Small (2402) | 46±1 | comparison only | | ||
| 543 | Minerva 62B | 46±2† | never released | | ||
| 544 | Phi-4-mini-reasoning 3.8B | 46±2 | api $0.13/M +local | ~4GB | | |
| 545 | DeepSeek-Coder-V2 Lite Instruct (16B-A2.4B) | 46±1 | local | ~9.6GB | +1 | |
| 546 | Qwen2-Math 1.5B Instruct | 46±2† | local | ~2.4GB | | |
| 547 | Llama 3.1 8B Instruct | 46±0 | api $0.22/M +local | ~6.2GB | ~47 tok/s | +25 |
| 548 | Ministral 3 3B | 46±1 | api $0.10/M +local | ~3.4GB | ~128 tok/s | +2 |
| 549 | Olmo 3 7B Instruct | 46±1 | local | ~5.8GB | | |
| 550 | GPT-3.5 text-davinci-003 | 46±1 | comparison only | +34 | ||
| 551 | Command R Plus (04-2024) | 46±1 | local | ~43GB | +6 | |
| 552 | Mistral NeMo Instruct (2407) | 46±1 | local | ~8.8GB | +9 | |
| 553 | GPT-4o Audio Preview (2024-10-01) | 46±2 | comparison only | | ||
| 554 | Llama 3.1 Tulu 3 8B | 46±1 | local | ~6.2GB | | |
| 555 | Phi-3-mini-4k-instruct (June 2024 update) | 46±1 | local | ~5.7GB | | |
| 556 | Yi 1.5 9B Chat | 46±1 | local | ~6.8GB | +1 | |
| 557 | Xwin-LM 70B v0.1 | 46±1 | local | ~28GB | | |
| 558 | Qwen1.5 32B Chat | 46±1 | local | ~22GB | | |
| 559 |  Eurus-2-7B-PRIME | 46±2 | local | ~6.4GB | | |
| 560 | Granite 20B Code Instruct (8K) | 46±1 | local | ~14GB | | |
| 561 | Qwen2 7B Instruct | 46±1 | local | ~5.4GB | +4 | |
| 562 | Command R7B (12-2024) | 46±1 | local | ~6.2GB | | |
| 563 |  Dolphin 2.6 Mixtral 8x7B | 46±1 | local | ~29GB | | |
| 564 | Gemma 3 4B | 45±1 | api $0.063/M +local | ~3.8GB | ~74 tok/s | +11 |
| 565 | Granite 3.0 8B Instruct | 45±1 | local | ~6.6GB | | |
| 566 | Granite 8B Code Instruct (4K) | 45±1 | local | ~6.6GB | | |
| 567 | Phi-4-mini 3.8B | 45±1 | api $0.13/M +local | ~4GB | +4 | |
| 568 | Ministral 8B (2410) | 45±1 | local | ~6.2GB | +6 | |
| 569 |  LLaVA-OneVision Qwen2 72B (ov-sft) | 45±3 | local | ~155GB | | |
| 570 | Code Llama 70B Python | 45±1 | local | ~28GB | | |
| 571 | Falcon3 7B Instruct | 45±2 | local | ~5.9GB | | |
| 572 | Code Tulu 2 34B | 45±2† | local | ~72GB | | |
| 573 | Granite 3.2 8B Instruct | 45±1 | local | ~6.6GB | | |
| 574 | OpenThinker 7B | 45±2† | local | ~6GB | | |
| 575 | Falcon3 10B Instruct | 45±2 | local | ~7.6GB | ~13 tok/s | |
| 576 | Yi 34B Chat | 45±1 | local | ~23GB | +1 | |
| 577 | Hermes 2 Theta Llama 3 8B | 45±2 | local | ~6.6GB | | |
| 578 | Llama 3.1 8B (base) | 45±3 | local | ~6.2GB | | |
| 579 | Gemma 3n E2B IT | 45±2 | local | ~4.1GB | | |
| 580 | Phi-3.5-mini-instruct | 45±1 | local | ~5.7GB | +4 | |
| 581 | Code Llama 70B (base) | 45±1 | local | ~28GB | | |
| 582 | Qwen2.5-VL 3B Instruct | 45±1 | local | ~2.8GB | | |
| 583 | LFM2.5 1.2B Instruct | 45±2 | local | ~1.3GB | | |
| 584 | Qwen3 1.7B | 45±1 | local | ~2.6GB | ~96 tok/s | +11 |
| 585 | Qwen3-VL 2B Instruct | 45±1 | local | ~2.7GB | | |
| 586 | Granite 3.1 8B Instruct | 45±1 | local | ~6.7GB | | |
| 587 | Reka Flash 21B (2024-02-26) | 45±1 | comparison only | | ||
| 588 | GPT-3.5 text-davinci-002 | 45±0 | comparison only | +22 | ||
| 589 | Claude Instant 1.1 | 45±2 | comparison only | | ||
| 590 | Qwen2.5 3B Instruct | 44±1 | local | ~2.6GB | ~61 tok/s | +5 |
| 591 | Flan-PaLM 62B | 44±1† | never released | | ||
| 592 | CogVLM2 Llama3 Chat 19B | 44±3 | local | ~43GB | | |
| 593 | Llama 3.2 11B Vision Instruct | 44±2 | local | ~8.9GB | | |
| 594 | LFM2 2.6B | 44±1 | local | ~2GB | | |
| 595 | Qwen2-Audio 7B Instruct | 44±3 | local | ~6.5GB | | |
| 596 | Tulu 2 DPO 70B | 44±1 | local | ~32GB | +1 | |
| 597 | Phi-3-small-8k-instruct | 44±2 | local | ~17GB | | |
| 598 | MiniCPM-V 2.6 | 44±2 | local | ~5.8GB | | |
| 599 | Yi-1.5 9B (base) | 44±2 | local | ~6.8GB | | |
| 600 | Claude Instant 1.2 | 44±1 | comparison only | | ||
| 601 | CodeGemma 7B IT (v1.0) | 44±1 | local | ~7.1GB | | |
| 602 | DeepSeek-Coder 1.3B Instruct | 44±1 | local | ~2.9GB | | |
| 603 | Aya 23 8B | 44±3 | local | ~6.2GB | | |
| 604 | Qwen2.5 Coder 1.5B Instruct | 44±1 | local | ~1.8GB | | |
| 605 | WizardLM 70B V1.0 | 44±1 | local | ~32GB | | |
| 606 | Mathstral 7B v0.1 | 44±1 | local | ~5.6GB | | |
| 607 | SmolLM3 3B | 44±1 | local | ~2.9GB | | |
| 608 | Llama 3 8B Instruct | 44±0 | local | ~6.6GB | +19 | |
| 609 | Hermes 3 Llama 3.1 8B | 44±1 | local | ~6.6GB | | |
| 610 | Command R (03-2024) | 44±1 | local | ~33GB | +6 | |
| 611 | Aya Vision 32B | 44±3 | api price unknown +local | ~71GB | | |
| 612 | WizardCoder Python 13B V1.0 | 44±2 | local | ~15GB | | |
| 613 | OpenChat 3.5 | 43±1 | local | ~6.1GB | | |
| 614 | Granite 34B Code Base (8K) | 43±1 | local | ~24GB | | |
| 615 | speechless-starcoder2-7b | 43±1 | local | ~6.1GB | | |
| 616 | Falcon2 11B (base) | 43±3 | local | ~8.2GB | | |
| 617 | OLMo 2 7B Instruct (1124) | 43±1 | local | ~5.8GB | | |
| 618 | Ministral 3B (2410) | 43±1 | comparison only | | ||
| 619 |  Snowflake Arctic Instruct | 43±2 | local | ~1007GB | | |
| 620 | Phi-3-vision-128k-instruct | 43±2 | local | ~10GB | | |
| 621 | WizardCoder-15B-V1.0 | 43±1 | local | ~12GB | | |
| 622 | cont-PaLM 62B (PaLM 62B trained to 1.3T tokens) | 43±1† | never released | | ||
| 623 | DeepSeek LLM 67B Chat | 43±1 | local | ~31GB | | |
| 624 | Qwen1.5 14B Chat | 43±1 | local | ~11GB | +3 | |
| 625 | Llama 2 70B (base) | 43±1 | local | ~32GB | | |
| 626 | Mixtral 8x7B v0.1 (base) | 43±1 | local | ~29GB | | |
| 627 | Code Llama 34B Python | 43±1 | local | ~23GB | | |
| 628 | speechless-coding-7b-16k-tora | 43±1 | local | ~16GB | | |
| 629 | InternVL2-4B | 43±2 | local | ~10GB | | |
| 630 | DeepSeek-Coder 33B Base | 43±1 | local | ~22GB | | |
| 631 | InternLM2 Chat 20B | 43±1 | local | ~14GB | | |
| 632 | CodeQwen1.5 7B (base) | 42±1 | local | ~6.4GB | | |
| 633 | Mixtral 8x7B Instruct v0.1 | 42±0 | local | ~28GB | +15 | |
| 634 | Nous Hermes 2 Mistral 7B DPO | 42±3 | local | ~6.1GB | | |
| 635 | Granite 4.0 1B | 42±1 | local | ~2.4GB | | |
| 636 | U-PaLM 62B | 42±2† | never released | | ||
| 637 | OpenHermes-2.5-Code-290k-13B | 42±1 | local | ~9.8GB | | |
| 638 | Granite 4.0 H 1B | 42±1 | local | ~2.3GB | | |
| 639 | Nous Hermes 2 Mixtral 8x7B DPO | 42±2 | local | ~30GB | | |
| 640 |  Deita 7B v1.0 | 42±2 | local | ~32GB | | |
| 641 | Molmo 7B-D (0924) | 42±1 | local | ~35GB | | |
| 642 | North Micro Vision Instruct (2.4B) | 42±2† | local | ~3.8GB | | |
| 643 |  Vicuna 13B v1.1 | 42±2 | comparison only | | ||
| 644 | OpenHermes 2.5 Mistral 7B | 42±1 | local | ~5.9GB | +2 | |
| 645 | Code Llama 34B Instruct | 42±1 | local | ~22GB | +1 | |
| 646 | Starling-LM 7B beta | 42±1 | local | ~6GB | +3 | |
| 647 | Code Llama 34B (base) | 42±1 | local | ~23GB | +1 | |
| 648 | Vicuna 33B v1.3 | 41±1 | local | ~23GB | +2 | |
| 649 | Phi-3-mini-128k-instruct (Apr 2024) | 41±2 | local | ~9.5GB | +1 | |
| 650 | Bielik 1.5B v3.0 Instruct | 41±3 | local | ~2.4GB | | |
| 651 | Code Tulu 2 13B | 41±2† | local | ~29GB | | |
| 652 |  Starling-LM 7B alpha | 41±1 | local | ~6.1GB | +2 | |
| 653 |  LLaVA-v1.6-34B | 41±2 | local | ~24GB | | |
| 654 | Granite 3.3 2B Instruct | 41±1† | local | ~3GB | | |
| 655 |  Llama3-OpenBioLLM-8B | 41±2 | local | ~6.6GB | | |
| 656 | Mistral 7B Instruct v0.3 | 41±1 | local | ~5.6GB | +4 | |
| 657 | LFM2 1.2B | 41±1 | local | ~1.3GB | | |
| 658 | DeepSeek-Coder 6.7B Base | 41±1 | local | ~8.8GB | | |
| 659 | Llama 2 70B Chat | 41±1 | local | ~32GB | +13 | |
| 660 | Granite 3.0 2B Instruct | 41±1 | local | ~3.2GB | | |
| 661 | Granite 8B Code Base (4K) | 41±1 | local | ~6.6GB | | |
| 662 | StarCoder2 15B | 41±1 | local | ~11GB | | |
| 663 | Llama 3.2 3B Instruct | 41±1 | api $0.12/M +local | ~3.3GB | ~139 tok/s | +11 |
| 664 | Falcon-H1 3B Instruct | 41±3 | local | ~3.4GB | | |
| 665 | Phi-2 (2.7B) | 40±1 | local | ~4.7GB | +2 | |
| 666 | speechless-code-mistral-7b-v1.0 | 40±1 | local | ~17GB | | |
| 667 | Llama-3-Instruct-8B-SimPO-v0.2 | 40±1 | local | ~6.6GB | | |
| 668 | Qwen-14B-Chat | 40±1 | local | ~11GB | | |
| 669 | Code Llama 13B Instruct | 40±1 | local | ~15GB | | |
| 670 | Tulu 2 13B | 40±2 | local | ~9.8GB | | |
| 671 |  MistralHermes-CodePro-7B-v1 | 40±1 | local | ~6.1GB | | |
| 672 | Falcon-180B-Chat | 40±2 | local | ~115GB | | |
| 673 | LLaMA 65B | 40±1 | comparison only | +7 | ||
| 674 | Phi-3-mini-4k-instruct (Apr 2024) | 40±2 | local | ~6GB | | |
| 675 | CodeGemma 7B (v1.0) | 40±1 | local | ~7.1GB | | |
| 676 | Code Llama 13B Python | 40±1 | local | ~9.8GB | | |
| 677 |  Apertus 8B Instruct (2509) | 40±3 | local | ~6.3GB | | |
| 678 | DeepSeek-V2 Lite Chat (16B-A2.4B) | 40±1 | local | ~11GB | | |
| 679 | Granite 3B Code Instruct (2K) | 40±1 | local | ~3.7GB | | |
| 680 | Tulu 2 DPO 13B | 40±1 | local | ~9.8GB | | |
| 681 |  BioMistral-7B | 39±3 | local | ~6.1GB | | |
| 682 | PaLM 62B | 39±1† | never released | +4 | ||
| 683 | Code Tulu 2 7B | 39±2† | local | ~16GB | | |
| 684 | SOLAR 10.7B Instruct v1.0 | 39±1 | local | ~8.3GB | +2 | |
| 685 | Gemma 2 2B IT | 39±1 | local | ~3.2GB | +4 | |
| 686 | Yi-1.5 6B (base) | 39±3 | local | ~5.1GB | | |
| 687 | Flan-T5 XXL | 39±1 | local | ~49GB | +12 | |
| 688 | Dolphin 2.2.1 Mistral 7B | 39±2 | local | ~5.9GB | | |
| 689 | Gemma 1.1 7B IT | 39±1 | local | ~7.1GB | +2 | |
| 690 | Phi-3.5-vision-instruct | 39±2 | local | ~10GB | | |
| 691 | Granite 20B Code Base (8K) | 39±1 | local | ~15GB | | |
| 692 | Granite 3.1 2B Instruct | 39±1 | local | ~3.1GB | | |
| 693 | Qwen1.5 7B Chat | 39±1 | local | ~9.7GB | +4 | |
| 694 | Code Llama 7B Python | 39±1 | local | ~5.8GB | | |
| 695 | Granite 3.2 2B Instruct | 39±1 | local | ~3GB | | |
| 696 | Falcon3 3B Instruct | 39±3 | local | ~3.4GB | | |
| 697 | Qwen2.5 Coder 0.5B Instruct | 39±1 | local | ~1GB | | |
| 698 | Vicuna 13B v1.5 | 38±1 | local | ~9.8GB | | |
| 699 | Zephyr 7B alpha | 38±1 | local | ~6.1GB | | |
| 700 | Code Llama 13B (base) | 38±1 | local | ~9.2GB | +1 | |
| 701 | Yi 1.5 6B Chat | 38±2 | local | ~4.4GB | | |
| 702 | Qwen2-VL 2B Instruct | 38±1 | local | ~2.1GB | | |
| 703 | WizardLM 13B V1.2 | 38±1 | local | ~9.8GB | | |
| 704 | Qwen2.5 1.5B Instruct | 38±1 | local | ~1.8GB | ~98 tok/s | +6 |
| 705 | Code Llama 7B Instruct | 38±1 | local | ~5.5GB | | |
| 706 | Tulu 2 7B | 38±2 | local | ~5.8GB | | |
| 707 | LFM2 700M | 38±1 | local | ~1GB | | |
| 708 | Qwen3.5 0.8B | 38±1 | local | ~1.4GB | ~94 tok/s | +2 |
| 709 |  MPT-30B-Chat | 38±2 | comparison only | | ||
| 710 | Mistral 7B Instruct v0.2 | 38±1 | local | ~5.8GB | +9 | |
| 711 | EXAONE 3.5 2.4B Instruct | 38±2 | local | ~3GB | | |
| 712 | LLaMA 33B | 38±1 | comparison only | +9 | ||
| 713 | Llama 3 8B (base) | 37±1 | local | ~6.6GB | | |
| 714 |  xDAN-L1-Chat-RL-v1 | 37±1 | local | ~6.1GB | | |
| 715 | Flan-PaLM 8B | 37±1† | never released | | ||
| 716 | Mistral 7B v0.1 (base) | 37±1 | local | ~5.8GB | +4 | |
| 717 | Flan-T5 XL | 37±1 | local | ~14GB | +8 | |
| 718 | Gemma 7B (base) | 37±1 | local | ~7.1GB | +3 | |
| 719 | Tulu 2 DPO 7B | 37±1 | local | ~5.8GB | | |
| 720 | Vicuna 13B v1.3 | 37±2 | local | ~29GB | | |
| 721 | Granite 3.1 3B-A800M Instruct | 37±2† | local | ~3.6GB | ~23 tok/s | |
| 722 | Granite 3.0 3B-A800M Instruct | 37±1 | local | ~3.7GB | | |
| 723 | DeepSeek R1 Distill Qwen 1.5B | 37±2 | local | ~1.9GB | +1 | |
| 724 | StarCoder2 7B | 37±1 | local | ~5.4GB | | |
| 725 | DeepSeek LLM 7B Chat | 37±2† | local | ~5.7GB | | |
| 726 | Vicuna 7B v1.1 | 37±2 | comparison only | | ||
| 727 | Falcon 40B (base) | 36±1 | local | ~27GB | +2 | |
| 728 | StarCoder 15.5B | 36±1 | local | ~11GB | | |
| 729 | Eurus-2-7B-SFT | 36±3† | local | ~6.4GB | | |
| 730 | Nous-Hermes-13b | 36±2 | local | ~9.8GB | | |
| 731 | Qwen3 0.6B | 36±1 | local | ~1.9GB | ~182 tok/s | +5 |
| 732 | PaLM 2 Chat (chat-bison@001) | 36±2 | comparison only | | ||
| 733 | Code Llama 7B (base) | 36±1 | local | ~5.5GB | +2 | |
| 734 | Llama 2 13B (base) | 36±1 | local | ~9.3GB | +4 | |
| 735 | Llama 2 13B Chat | 36±1 | local | ~15GB | +10 | |
| 736 | Zephyr 7B beta | 36±1 | local | ~6.1GB | +3 | |
| 737 |  Guanaco 33B | 35±1 | local | ~70GB | | |
| 738 | StarCoder2 3B | 35±1 | local | ~3.4GB | | |
| 739 | Vicuna 7B v1.5 | 35±1 | local | ~5.8GB | +1 | |
| 740 | Granite 3B Code Base (2K) | 35±1 | local | ~3.7GB | | |
| 741 | StarCoderBase 15.5B | 35±1 | local | ~12GB | | |
| 742 | OLMo 2 1B Instruct (0425) | 35±1 | local | ~2.4GB | | |
| 743 | ChatGLM3 6B 128K | 35±4 | local | ~5.8GB | | |
| 744 | Minerva 8B | 35±2† | never released | | ||
| 745 | Vicuna 7B v1.3 | 35±2 | local | ~5.8GB | | |
| 746 |  Stable Code 3B | 35±1 | local | ~4.8GB | | |
| 747 | Orca 2 13B | 35±3 | local | ~9.8GB | | |
| 748 |  T0 11B | 35±3 | comparison only | | ||
| 749 | Qwen1.5 4B Chat | 34±2 | local | ~4.1GB | | |
| 750 | DeepSeek-Coder 1.3B Base | 34±1 | local | ~2.4GB | | |
| 751 | Gemma 7B IT | 34±1 | local | ~7.1GB | +5 | |
| 752 | Mistral 7B Instruct v0.1 | 34±1 | local | ~5.9GB | +3 | |
| 753 | LaMDA 137B (pre-trained, LaMDA-PT) | 34±1† | never released | | ||
| 754 | GPT-3 text-davinci-001 | 34±1 | comparison only | +5 | ||
| 755 | CodeGemma 2B (v1.0) | 34±1 | local | ~3.2GB | | |
| 756 | Qwen2 1.5B Instruct | 34±1 | local | ~1.8GB | +1 | |
| 757 |  Ghost 7B Alpha | 34±2 | local | ~6.1GB | | |
| 758 | Falcon-H1 1.5B Instruct | 33±4 | local | ~2.4GB | | |
| 759 | Gemma 3 1B | 33±1 | local | ~1.7GB | ~98 tok/s | +7 |
| 760 | SmolLM2 1.7B Instruct | 33±1 | local | ~3GB | | |
| 761 | Qwen-7B (base) | 33±1 | local | ~6.6GB | | |
| 762 | LFM2 350M | 33±2 | local | ~1.7GB | | |
| 763 | OPT-175B | 33±1 | comparison only | | ||
| 764 | BLOOMZ 176B | 33±2 | comparison only | | ||
| 765 | StarCoderBase 7B | 33±1 | local | ~6GB | | |
| 766 | Codex code-davinci-001 | 33±1 | comparison only | | ||
| 767 | Llama 3.2 1B Instruct | 33±1 | api $0.020/M +local | ~1.6GB | ~28 tok/s | +13 |
| 768 | Phi-1.5 (1.3B) | 33±1 | local | ~2.4GB | | |
| 769 | LLaMA 13B | 33±1 | comparison only | +19 | ||
| 770 | Qwen-7B-Chat | 32±1 | local | ~6.6GB | | |
| 771 | Llama 2 7B Chat | 32±1 | local | ~8.8GB | +13 | |
| 772 |  Koala 13B | 32±1 | comparison only | +1 | ||
| 773 | OLMo 7B Instruct | 32±2 | local | ~16GB | | |
| 774 | Llama 2 7B (base) | 32±1 | local | ~8.5GB | +6 | |
| 775 | GPT-3 davinci (API base model) | 32±1 | comparison only | +13 | ||
| 776 | Flan-T5 Large | 32±1 | local | ~4.8GB | +5 | |
| 777 | Granite 3.1 1B-A400M Instruct | 31±2† | local | ~2.3GB | ~20 tok/s | |
| 778 | Falcon 40B Instruct | 31±2 | local | ~27GB | | |
| 779 | Gemma 2B (base) | 31±1 | local | ~3.2GB | +3 | |
| 780 | Gemma 1.1 2B IT | 31±1 | local | ~3.2GB | +1 | |
| 781 | Falcon 7B (base) | 31±1 | local | ~5.9GB | +4 | |
| 782 |  Alpaca 7B | 31±1 | comparison only | | ||
| 783 |  Baichuan2-7B-Base | 30±1 | local | ~17GB | | |
| 784 | LLaMA 7B | 30±1 | comparison only | +17 | ||
| 785 | NexusRaven V2 13B | 30±3 | local | ~9.3GB | | |
| 786 | MPT-7B (base) | 30±1 | comparison only | | ||
| 787 | ChatGLM3-6B | 30±2 | local | ~5.8GB | | |
| 788 | Qwen2.5 0.5B Instruct | 30±2 | local | ~1GB | +1 | |
| 789 | OLMo 7B | 30±1† | local | ~30GB | | |
| 790 |  OpenLLaMA 7B v2 | 30±1 | local | ~9.1GB | | |
| 791 | StableLM-3B-4E1T | 30±1 | local | ~3.3GB | | |
| 792 | StarCoderBase 3B | 30±1 | local | ~3.4GB | | |
| 793 | Granite 3.0 1B-A400M Instruct | 30±1 | local | ~2.3GB | | |
| 794 | MPT-7B-Chat | 30±2 | comparison only | | ||
| 795 | StableLM-Base-Alpha 7B v2 | 30±1 | local | ~30GB | | |
| 796 | OPT-66B | 29±1 | local | ~141GB | +2 | |
| 797 |  GPT4All-13B-Snoozy | 29±2 | local | ~56GB | | |
| 798 | Alpaca 13B | 29±2 | comparison only | | ||
| 799 | EuroLLM 1.7B Instruct | 29±3 | local | ~2.4GB | | |
| 800 | Granite 4.0 H 350M | 29±1† | local | ~1.7GB | | |
| 801 | BLOOMZ 7.1B | 29±2 | comparison only | | ||
| 802 | Gemma 2B IT | 29±1 | local | ~3.1GB | +3 | |
| 803 | BLOOM 176B | 29±1 | local | ~372GB | +1 | |
| 804 | Granite 4.0 350M | 28±1 | local | ~1.7GB | | |
| 805 | OPT-30B | 28±1 | local | ~64GB | +5 | |
| 806 | UL2 20B | 28±3† | comparison only | | ||
| 807 |  GPT-NeoX-20B | 28±1 | local | ~14GB | +1 | |
| 808 | Qwen2 0.5B Instruct | 28±2 | local | ~1GB | | |
| 809 |  RedPajama-INCITE 7B Base | 28±1 | local | ~16GB | | |
| 810 | ChatGLM2-6B | 28±2 | local | ~15GB | | |
| 811 |  RWKV-4-Raven 14B | 28±2 | comparison only | | ||
| 812 | LaMDA 68B (pre-trained) | 27±1† | never released | | ||
| 813 | BLOOMZ 3B | 27±2 | comparison only | | ||
| 814 | Falcon3 1B Instruct | 26±4 | local | ~2.4GB | | |
| 815 | Qwen1.5 1.8B Chat | 26±2 | local | ~2.7GB | | |
| 816 | Tk-Instruct 11B (def) | 26±4 | comparison only | | ||
| 817 | OPT-13B (13 billion parameters) | 26±1 | local | ~29GB | +4 | |
| 818 | Falcon 7B Instruct | 26±2 | local | ~5.9GB | +1 | |
| 819 | PaLM 8B | 26±1† | never released | | ||
| 820 |  OpenAssistant SFT-4 Pythia 12B | 25±1 | local | ~9.5GB | +1 | |
| 821 | Flan-T5 Base | 25±1 | local | ~2.5GB | +3 | |
| 822 | RedPajama-INCITE Base 3B v1 | 25±3 | local | ~3.4GB | | |
| 823 | GPT-3 curie (API base model) | 25±1 | comparison only | +1 | ||
| 824 | Pythia 12B | 25±1 | local | ~9.5GB | +1 | |
| 825 | GPT-J 6B | 25±1 | local | ~14GB | +6 | |
| 826 | ChatGLM-6B | 25±1 | local | ~16GB | +3 | |
| 827 | OpenLLaMA 3B v2 | 25±1 | local | ~4.2GB | | |
| 828 | StableLM-Base-Alpha 3B v2 | 24±1 | local | ~13GB | | |
| 829 | OPT-6.7B | 24±1 | local | ~16GB | +5 | |
| 830 | T0 3B | 23±4 | comparison only | | ||
| 831 | Qwen1.5 0.5B Chat | 23±3 | local | ~1.7GB | | |
| 832 | BLOOMZ 1.7B | 23±3 | comparison only | | ||
| 833 | Pythia 6.9B | 23±1 | local | ~6.1GB | +2 | |
| 834 | T0++ (T0pp) 11B | 23±3 | comparison only | | ||
| 835 | Phi-1 (1.3B) | 22±4 | local | ~2.4GB | | |
| 836 | Dolly v2 12B | 22±1 | comparison only | +3 | ||
| 837 | SmolLM2 360M Instruct | 22±3 | local | ~1GB | | |
| 838 | FastChat-T5 3B | 21±2 | local | ~8.5GB | | |
| 839 | OLMo 1B | 21±2† | local | ~2.2GB | | |
| 840 | BLOOM 7.1B | 21±1 | local | ~7.1GB | +7 | |
| 841 | Pythia 2.8B | 20±1 | local | ~3.4GB | | |
| 842 | Pythia 2.8B (deduped) | 20±1 | local | ~3.4GB | | |
| 843 | Falcon-RW 1B | 20±2 | local | ~4.3GB | | |
| 844 | OPT-2.7B | 19±1 | local | ~7.1GB | +3 | |
| 845 | StableLM-Tuned-Alpha 7B | 18±2 | local | ~6.9GB | | |
| 846 | GPT-Neo 2.7B | 18±1 | local | ~13GB | +8 | |
| 847 | BLOOMZ 1.1B | 18±3 | comparison only | | ||
| 848 | T5 Large | 17±2 | local | ~4.6GB | | |
| 849 | Gemma 3 270M IT | 17±4 | local | ~1.1GB | | |
| 850 | Flan-T5 Small | 16±3 | local | ~1.8GB | | |
| 851 | BLOOM 3B | 16±1 | local | ~4GB | +2 | |
| 852 | GPT-3 babbage (API base model) | 15±1 | comparison only | | ||
| 853 | BLOOMZ 560M | 14±4 | comparison only | | ||
| 854 | OPT-1.3B | 14±1 | local | ~4.3GB | +5 | |
| 855 | T5 Base | 14±3 | local | ~2.4GB | | |
| 856 | Pythia 1.4B | 14±1 | local | ~2.4GB | | |
| 857 | LaMDA 8B (pre-trained) | 12±2† | never released | | ||
| 858 | Pythia 1B | 12±1 | local | ~2.2GB | | |
| 859 | GPT-3 text-curie-001 | 12±1 | comparison only | +3 | ||
| 860 | GPT-Neo 1.3B | 12±2 | local | ~7.1GB | +3 | |
| 861 | T5 Small | 12±3 | local | ~1.8GB | | |
| 862 | BLOOM 1.7B | 11±1 | local | ~3.1GB | +1 | |
| 863 |  Zyte-1B | 11±3 | local | ~2.2GB | | |
| 864 | LaMDA 2B (pre-trained) | 10±2† | never released | | ||
| 865 | StableLM-Base-Alpha 7B | 8±1 | local | ~35GB | | |
| 866 | Pythia 410M | 8±2 | local | ~1.8GB | | |
| 867 | GPT-3 ada (API base model) | 7±1 | comparison only | | ||
| 868 | BLOOM 1.1B | 7±1 | local | ~2.6GB | | |
| 869 | GPT-2 XL (1.5B) | 7±1 | local | ~2.8GB | ~75 tok/s | +9 |
| 870 | GPT-3 text-babbage-001 | 4±2 | comparison only | | ||
| 871 | GPT-3 text-ada-001 | 1±3 | comparison only | | ||
| 872 | OPT-350M | 1±1 | local | ~2.2GB | +1 | |
| 873 | LaMDA 422M (pre-trained) | 1±3† | never released | | ||
| 874 | StableLM-Base-Alpha 3B | 1±2 | local | ~17GB | | |
| 875 | BLOOM 560M | 0±1 | local | ~2.1GB | +1 | |
| 876 | Pythia 160M | -2±2 | local | ~1.6GB | | |
| 877 | GPT-2 Large (774M) | -3±1 | local | ~2GB | ~137 tok/s | +3 |
| 878 | OPT-125M | -3±1 | local | ~1.8GB | +1 | |
| 879 | GPT-Neo 125M | -3±2 | local | ~2.1GB | | |
| 880 | GPT-2 Medium (355M) | -5±1 | local | ~1.8GB | +3 | |
| 881 | Pythia 70M | -13±2 | local | ~1.6GB | | |
| 882 | GPT-2 Small (124M) | -13±2 | local | ~1.6GB | +3 | |
| — | Grok Build 0.1 | no reviewed results | api $1.25/M | |||
| — |  K2 Horizon 7B | not scored yet | local | ~8GB | ||
| — | LFM2.5 24B-A2B | not scored yet | local | ~17GB |
+ 306 models with fewer than 3 results left off, since one or two tests say little about overall ability · show them
Includes 231 comparison-only models with no verified API or download · hide them
885 of 1219 catalogue models listed here · sources and licences
Models outside this leaderboard (28 models)
No current run option verified: GPT-3 Small (125M, Brown et al. 2020 paper model) · GPT-3 Medium (350M, Brown et al. 2020 paper model) · GPT-3 Large (760M, Brown et al. 2020 paper model) · GPT-3 XL (1.3B, Brown et al. 2020 paper model) · GPT-3 2.7B (Brown et al. 2020 paper model) · GPT-3 6.7B (Brown et al. 2020 paper model) · GPT-3 13B (Brown et al. 2020 paper model) · GPT-3 175B (Brown et al. 2020 paper model) · Jurassic-1 Jumbo (J1-Jumbo, 178B) · Jurassic-1 Large (J1-Large, 7.5B) · Gopher (280B) · Chinchilla (70B) · GLaM 0.1B/64E (MoE) · GLaM 1.7B/64E (MoE) · GLaM 8B/64E (MoE) · GLaM (64B/64E, 1.2T MoE) · GLaM dense 0.1B · GLaM dense 1.7B · GLaM dense 8B · GLaM dense 137B · Luminous Base (13B) · Luminous Extended (30B) · Luminous Supreme (70B) · Gopher family 417M · Gopher family 1.4B · Gopher family 7.1B · Gopher family 44M · Gopher family 117M
* new or thinly tested: ranked on too few benchmark results · † only the maker’s own numbers so far · ~RAM is estimated; full context may need more memory