Tier List

AI model tier list for your hardware.

Every model graded S through F based on fit, speed, and headroom for RTX 4070 12GB running coding workloads.

HardwareWorkload

Qwen3.5 9B9B Meta Llama 3.1 8B Instruct8B Qwen3.5 9B Uncensored HauhauCS Aggressive9B llava llama 3 8b v1 18B DeepSeek R1 0528 Qwen3 8B8B Meta Llama 3 8B Instruct8B DeepSeek R1 Distill Llama 8B8B Qwen3.5 9B9B Dolphin3.0 Llama3.1 8B8B Llama 3 8B Instruct 32k v0.18B Meta Llama 3.1 8B Instruct8B vntl llama3 8b v28B DeepSeek R1 0528 Qwen3 8B8B DeepSeek R1 0528 Qwen3 8B8B Qwen3 8B DeepSeek v3.2 Speciale Distill8B Yi Coder 9B Chat9B glm 4 9b chat 1m9B Hermes 3 Llama 3.1 8B8B Aya Expanse 8B8B Hermes 2 Pro Llama 3 8B8B Dolphin3.0 Llama3.1 8B8B Dolphin 2.9 8B8B granite 8b code instruct 4k8B dolphin 2.9.4 llama3.1 8b8B Yi 1.5 9B Chat9B InternVL2 8B8B openchat 3.6 8b 20240522 IMat8B aya expanse 8b8B dolphin v2 8b abliterated i18B MiniCPM-V 2.6 8B8B Ministral 3 8B8B Nemotron Nano 8B8B EXAONE 3.5 7.8B Instruct7.800000190734863B aya expanse 8b orthogonal heretic i18B EXAONE 3.5 7.8B Instruct7.800000190734863B EXAONE 3.5 7.8B Instruct i17.800000190734863B Yi 9B Coder i19B aya expanse 8b orthogonal heretic8B exaone 3.0 7.8b it7.800000190734863B CodeGeeX 4 9B9B HelpingAI2 9B9B Nous Hermes 1.09B Yi Coder 9B9B DeepSeek R1 Distill 8B8B Qwen 3 8B8B internlm2 5 1 8b chat i18B internlm3 8b instruct abliterated i18B Llama 2 7B Chat7B HelpingAI2 9B i19B Granite Code 8B8B HelpingAI 9B 200k i19B GLM-4 9B9B Mistral 7B Instruct v0.27B HelpingAI 9B i19B Ministral 8B8B Mistral 7B Instruct v0.37B Granite 3.1 8B8B Gemma 2 9B9B Yi 1.5 9B9B Llama 3.1 8B8B StarCoder 7B7B Mistral 7B Instruct v0.37B Hermes 2 Pro Mistral 7B7B Zephyr 7B Beta7B zephyr 7B beta7B OpenChat 7B7B Nous Hermes 2 Mistral 7B DPO7B HelpingAI2.5 10B i110B Falcon H1R 7B7B falcon mamba 7b instruct Q4 K M7B Magistral 7B7B zephyr 7B alpha7B speechless zephyr code functionary 7b7B

StarCoder2 7B7B japanese stablelm instruct gamma 7B7B Falcon H1 7B Instruct7B Falcon H1R 7B7B DeepSeek R1 Distill 7B7B OpenHermes 2.5 7B7B Qwen 2.5 VL 7B7B Solar 7B7B SQLCoder 7B7B Mamba Codestral 7B v0.17B stabilityai japanese stablelm base gamma 7b7B Starling LM 7B7B baichuan2 7b chat7B DiscoPOP zephyr 7b gemma7B InternLM Chat 7B7B Mistral 7B Instruct v0.37B Neural Chat 7B7B Qwen 2.5 Math 7B7B starcoder2 7b7B zephyr 7b beta Mistral 7B Instruct v0.27B OpenChat 3.5 7B Qwen v2.0 i17B OpenChat 3.5 7B Starling v2.0 i17B internlm2 math plus 7b IMat7B InternLM 7B7B CodeNinja 1.0 OpenChat 7B i17B internlm2 5 7b chat i17B OpenSafetyLab MD Judge v0 2 internlm2 7b7B MD Judge v0 2 internlm2 7b i17B Baichuan 7B7B Codestral Mamba 7B7B DevStral 7B7B Qwen 2.5 Coder 7B7B zephyr 7b gemma sft african ultrachat 100k7B jointpreferences mistral 7b sft helpful7B zephyr 7b dpo full i17B blossom v3 baichuan2 7b i17B blossom v1 baichuan 7b i17B StarCoder2 7B7B LLaVA 1.5 7B7B Samantha 7B7B WizardMath 7B7B Qwen 2.5 7B7B CodeLlama 7B Instruct7B MPT-7B-Instruct7B DeepSeek LLM 7B7B Vicuna 7B7B Falcon 7B Instruct7B Yi 1.5 6B Chat6B stablelm 2 zephyr 1 6b6B Yi 1.5 6B Chat6B stablelm 2 1 6b chat imatrix6B HelpingAI2 6B6B HelpingAI2 6B i16B Yi 1.5 6B6B Llama 3.2 11B Vision11B SOLAR 10.7B Instruct v1.0 uncensored10.699999809265137B Nous Hermes 2 SOLAR 10.7B10.699999809265137B gemma 3 12b it12B solar finalised finetuned Model 10.7B i110.699999809265137B Helply 10.2b chat i110.199999809265137B SOLAR 10.7B v1.010.699999809265137B HelpingAI2.5 5B i15B Pixtral 12B12B Gemma 3 12B12B Qwen3.5 4B4B Mistral Nemo 12B12B gemma 3 4b it4B LLaVA 1.6 13B13B gemma 3 4b it4B CodeLlama 13B Instruct13B OLMo 2 13B13B Llama 3.2 3B Instruct3B Vicuna 13B13B WizardLM 13B13B StableLM 2 12B12B Phi 4 Mini 4B4B Gemma 3 4B4B Phi 3.5 Mini 4B4B Qwen2.5 3B Instruct3B Nemotron Mini 4B4B Qwen 3 4B4B Llama 3.2 3B Instruct3B Ministral 3 3B Instruct 25123B Phi 3 Mini 3.8B3.799999952316284B gemma 2 2b it2B HELVETE 3B3B Hermes 3 Llama 3.2 3B3B stablelm zephyr 3b3B StarCoder2 3B3B AI21 Jamba Reasoning 3B3B stablelm 3b 4e1t3B ai21labs AI21 Jamba Reasoning 3B3B ai21labs AI21 Jamba2 3B3B SmolLM3 3B3B HelpingAI 3B hindi i13B AI21 Jamba2 3B3B HelpingAI 3B hindi3B AI21 Jamba2 3B i13B Ministral 3 3B3B gemma 2b2B EXAONE 3.5 2.4B Instruct2.4000000953674316B Gemmasutra Mini 2B v12B gemma 2 2b it2B Llama 3.2 3B3B Qwen2.5 1.5B Instruct1.5B DeepSeek R1 Distill Qwen 1.5B1.5B Llama 3.2 1B Instruct Q8 01B Yi Coder 1.5B Chat1.5B TinyLlama 1.1B Chat v1.01.100000023841858B stablelm 2 zephyr 1.6b1.600000023841858B logos16v2 stablelm2 1.6b i11.600000023841858B Yi Coder 1.5B1.5B TinyLlama 1.1B Chat v0.31.100000023841858B Falcon H1 1.5B Instruct1.5B TinyLlama 1.1B Chat v0.61.100000023841858B Llama 3.2 1B Instruct1B gemma 3 1b it1B Gemma 2 2B2B EXAONE 4.0 1.2B1.2000000476837158B SmolVLM 500M Instruct0.5B DeepSeek R1 1.5B1.5B TinyLlama 1.1B Chat v1.0 imatrix1.100000023841858B Qwen 2.5 Coder 1.5B1.5B Falcon3 1B Instruct abliterated1B Jina Embeddings v30.5720000267028809B embeddinggemma 300M0.30000001192092896B BGE M30.5680000185966492B Qwen 3 1.7B1.7000000476837158B mxbai Embed Large0.33500000834465027B Snowflake Arctic Embed L0.33500000834465027B TinyLlama 1.1B1.100000023841858B BGE Large EN v1.50.33500000834465027B granite embedding 107m multilingual0.10700000077486038B Nomic Embed Text v1.50.13699999451637268B Falcon H1 Tiny 90M Instruct0.09000000357627869B Llama 3.2 1B1B Gemma 3 1B1B All MiniLM L6 v20.023000000044703484B Qwen 3 0.6B0.6000000238418579B DeepSeek Coder V2 16B16B DeepSeek R1 Distill Qwen 14B14B Ministral 3 14B14B DeepSeek R1 Distill 14B14B Qwen 2.5 Coder 14B14B NousResearch Hermes 4 14B14B Phi-4-reasoning-plus 14B14.699999809265137B Qwen 3 14B14B Phi 3 Medium 14B14B Phi-4 14B14B Qwen 2.5 14B14B GGUF SOLARized GraniStral 14B 2102 YeAM HCT 32QKV14B internlm JanusCoder 14B14B GGUF SOLARized GraniStral 14B 1902 YeAM HCT14B Phi 4 reasoning vision 15B15B StarCoder2 15B15B starcoder2 15b instruct v0.115B starcoder2 15b i115B starcoder2 15b instruct v0.115B Baichuan 13B13B HelpingAI 15B i115B Nous Dolphin 13B13B Cerebras-GPT 13B13B

Codestral 2 25.0822B Codestral 22B22B CogVLM2 19B19B Command A 111B111B Command R 35B35B Command R+ 104B104B DeepSeek Coder V2 236B236B DeepSeek LLM 67B67B DeepSeek R1 671B671B DeepSeek R1 Distill 32B32B DeepSeek V2.5 236B236B DeepSeek V3 671B671B Devstral 2 123B Instruct123B Devstral Small 2 24B Instruct24B Devstral Small 1.124B Falcon 40B Instruct40B Gemma 2 27B27B Gemma 3 27B27B GLM-5744B GPT-OSS 20B21B Granite Code 20B20B Solar Open 100B100B Baichuan M2 32B Q4 K M32B baichuan inc Baichuan M2 32B32B Codestral 22B v0.122B cognitivecomputations Dolphin Mistral 24B Venice Edition24B cognitivecomputations Dolphin3.0 R1 Mistral 24B24B internlm2 5 20b chat20B Qwen3 48B A4B Savant Commander Distill 12X Closed Open Heretic Uncensored48B internlm2 limarp chat 20b20B Codestral 22B v0.1 IMat22B EXAONE 4.0 32B32B K EXAONE 236B A23B236B Codestral 22B v0.122B Qwen3.5 35B A3B35B gemma 3 27b it27B Llama 3.3 70B Instruct70B Mistral Small 24B Instruct 250124B mistral small 3.1 24b instruct 2503 hf24B aya expanse 32b heretic MPOA i132B Baichuan M3 235B235B Baichuan M3 235B i1235B BaichuanMed OCR 72B i172B Codestral 21B Pruned i121B Codestral 22B v0.1 i122B Codestral RAG 19B Pruned i119B Dolphin Mistral GLM 4.7 Flash 24B Venice Edition Thinking Uncensored i124B internlm2 math plus 20b i120B Solar Open 100B i1100B Solar Open 69B REAP i169B Hermes 4.3 36B36B stabilityai japanese stablelm instruct beta 70b70B Codestral 22B v0.122B gemma 3 27b it27B Mistral Small 3.2 24B Instruct 250624B Qwen3.5 122B A10B122B Qwen3.5 27B27B Qwen3.5 35B A3B35B Qwen3.5 397B A17B397B cognitivecomputations Dolphin Mistral 24B Venice Edition24B InternLM 20B20B Kimi K2.51000B Llama 3.1 70B70B Llama 3.3 70B70B Llama 4 Maverick 17B 128E400B Llama 4 Scout 17B 16E109B Magistral Small 250724B Mistral Large 3675B Mistral Small 24B24B Mistral Small 3.2 24B24B Mistral Small 4 119B119B Mixtral 8x22B141B Mixtral 8x7B47B MPT-30B-Instruct30B Nemotron 70B70B OLMo 2 32B32B Pixtral Large 124B124B Qwen 2.5 32B32B Qwen 2.5 72B72B Qwen 2.5 Coder 32B32B Qwen 2.5 Math 72B72B Qwen 2.5 VL 72B72B Qwen 3 235B A22B235B Qwen 3 30B A3B30.5B Qwen 3 32B32B Qwen3-Coder 30B A3B Instruct30.5B Qwen3-Coder 480B A35B Instruct480B Qwen3-Coder-Next80B Qwen3-VL 30B A3B Instruct30B StarCoder 15B15B StarCoder2 15B15B Yi 1.5 34B34B Yi 34B Chat34B