New open models
Every open model released in the last twelve months, newest first, with its intelligence rank and the memory it needs.
Follow new releases (RSS)September 2026 · 1
| Model | Parameters | ECI | Memory |
|---|---|---|---|
 DeepSeek V4.1 FlashDeepSeek | 763.2B · 16B active | 154.9 #3 | 436 GB |
August 2026 · 8
| Model | Parameters | ECI | Memory |
|---|---|---|---|
| DeepSeek V4 Pro 0813DeepSeek | 1.7T · 49B active | 155.3 #2 | 887 GB |
 GLM 5.3 FlashZ.ai | 321.3B · 18B active | 151.9 #5 | 176 GB |
 Qwen3.8 27BQwen | 27.8B | 149.4 #11 | 17 GB |
| Qwen3.8 Flash NextQwen | 180B · 6B active | — | 112 GB |
 Granite 4.2 3BIBM | 3.7B | — | 3 GB |
| Granite 4.2 8BIBM | 8.8B | — | 7 GB |
| Granite 4.2 30BIBM | 29.3B | — | 19 GB |
 Nemotron 3.5 Lightning 30B A3BNVIDIA | 31.6B · 3B active | — | 24 GB |
July 2026 · 5
| Model | Parameters | ECI | Memory |
|---|---|---|---|
 Kimi K3Moonshot AI | 2.8T · 104B active | 157.5 #1 | 1,584 GB |
| DeepSeek V4 Flash 0731DeepSeek | 304.2B · 13B active | 154.5 #4 | 161 GB |
| TInkling SmallThinking Machines Lab | 266B · 12B active | 150.2 #8 | 149 GB |
| TInklingThinking Machines Lab | 952.4B · 41B active | 148.5 #13 | 534 GB |
 LFM2.5 2.6BLiquid AI | 2.7B | — | 2 GB |
June 2026 · 5
| Model | Parameters | ECI | Memory |
|---|---|---|---|
| GLM 5.2Z.ai | 753.3B · 40B active | 151.8 #6 | 424 GB |
| Kimi K2.7 CodeMoonshot AI | 1T · 32B active | 150 #9 | 579 GB |
 MiniMax M3MiniMax | 427B · 23B active | 147 #15 | 245 GB |
| Nemotron 3 Ultra 550B A55BNVIDIA | 560.5B · 55B active | 146.2 #20 | 322 GB |
 North Mini Code 1.0Cohere | 30.5B · 3B active | — | 18 GB |
May 2026 · 2
| Model | Parameters | ECI | Memory |
|---|---|---|---|
 Gemma 4 12BGoogle | 12B | — | 7 GB |
| Command A+Cohere | 218.8B · 25B active | — | 127 GB |
April 2026 · 12
| Model | Parameters | ECI | Memory |
|---|---|---|---|
| Kimi K2.6Moonshot AI | 1T · 32B active | 151.1 #7 | 579 GB |
| GLM 5.1Z.ai | 753.9B · 40B active | 149.8 #10 | 429 GB |
| DeepSeek V4 ProDeepSeek | 1.6T · 49B active | 149.1 #12 | 887 GB |
| Qwen3.6 27BQwen | 27.8B | 146.5 #18 | 16 GB |
| DeepSeek V4 FlashDeepSeek | 290.9B · 13B active | 146.1 #21 | 167 GB |
| MiniMax M2.7MiniMax | 228.7B · 10B active | 145.9 #23 | 132 GB |
| Qwen3.6 35B A3BQwen | 36B · 3B active | 143.9 #26 | 21 GB |
| Gemma 4 31BGoogle | 31.3B | 142.7 #29 | 19 GB |
| Gemma 4 26B A4BGoogle | 25.8B · 3.8B active | 141.9 #31 | 17 GB |
 Mistral Medium 3.5 128BMistral AI | 127.7B | 141.4 #32 | 73 GB |
| Gemma 4 E4BGoogle | 8B | — | 5 GB |
| Gemma 4 E2BGoogle | 5.1B | — | 3 GB |
March 2026 · 1
| Model | Parameters | ECI | Memory |
|---|---|---|---|
| Mistral Small 4 119BMistral AI | 119.4B · 6.5B active | — | 68 GB |
February 2026 · 9
| Model | Parameters | ECI | Memory |
|---|---|---|---|
| MiniMax M2.5MiniMax | 228.7B · 10B active | 146.7 #16 | 131 GB |
| Qwen3.5 397B A17BQwen | 403.4B · 17B active | 146.7 #17 | 228 GB |
| GLM 5Z.ai | 753.9B · 40B active | 145.8 #24 | 426 GB |
| Qwen3.5 35B A3BQwen | 36B · 3B active | 142.5 #30 | 21 GB |
| Qwen3.5 9BQwen | 9.7B | 139.5 #37 | 6 GB |
| Qwen3.5 4BQwen | 4.7B | — | 3 GB |
| Qwen3.5 2BQwen | 2.3B | — | 2 GB |
| Qwen3.5 0.8BQwen | 0.9B | — | 1 GB |
| Qwen3.5 122B A10BQwen | 125.1B · 10B active | — | 72 GB |
January 2026 · 3
| Model | Parameters | ECI | Memory |
|---|---|---|---|
| Kimi K2.5Moonshot AI | 1T · 32B active | 148 #14 | 579 GB |
| Qwen3 Coder NextQwen | 79.7B · 3B active | — | 46 GB |
| GLM 4.7 FlashZ.ai | 31.2B · 3B active | — | 18 GB |
December 2025 · 7
| Model | Parameters | ECI | Memory |
|---|---|---|---|
| DeepSeek V3.2DeepSeek | 685.4B · 37B active | 146.3 #19 | 379 GB |
| GLM 4.7Z.ai | 358.3B · 32B active | 143.5 #28 | 205 GB |
| Ministral 3 14BMistral AI | 14B | — | 9 GB |
| Ministral 3 8BMistral AI | 8.9B | — | 6 GB |
| Ministral 3 3BMistral AI | 3.9B | — | 3 GB |
| Devstral Small 2 24BMistral AI | 24B | — | 15 GB |
 Olmo 3.1 32B ThinkAllen Institute for AI | 32.2B | — | 20 GB |
November 2025 · 2
| Model | Parameters | ECI | Memory |
|---|---|---|---|
| Kimi K2 ThinkingMoonshot AI | 1T · 32B active | 146 #22 | 579 GB |
| Olmo 3 7BAllen Institute for AI | 7.3B | — | 7 GB |
October 2025 · 1
| Model | Parameters | ECI | Memory |
|---|---|---|---|
| Kimi Linear 48B A3BMoonshot AI | 49.1B · 3B active | — | 28 GB |
Released is the month the weights appeared on Hugging Face. Downloads are Hugging Face’s count for the last 30 days, original weights plus the GGUF repo we link, so they include servers as well as people running models at home. Memory at Q4_K_M with an 8K context.
Older models, newest first