The following models are available to use with our fine-tuning API. Get started with fine-tuning a model!
  • Training Precision Type indicates the precision type used during training for each model.
    • AMP (Automated Mixed Precision): AMP allows the training speed to be faster with less memory usage while preserving convergence behavior compared to using float32. Learn more about AMP in this PyTorch blog.
    • bf16 (bfloat 16): This uses bf16 for all weights. Some large models on our platform use full bf16 training for better memory usage and training speed.
  • For batch sizes of 1, Gradient accumulation 8 is used, so effectively you will get batch size 8 (iteration time is slower).
  • Long-context fine-tuning of Llama 3.1 (8B) Reference, Llama 3.1 (70B) Reference, Llama 3.1 Instruct (70B) Reference for context sizes of 32K-131K is only supported using the LoRA method.
  • For Llama 3.1 (405B) Fine-tuning, please contact us.
Request a model

LoRA Fine-tuning

OrganizationModel NameModel String for APIContext Length (SFT)Context Length (DPO)Max Batch Size (SFT)Max Batch Size (DPO)Min Batch SizeTraining Precision Type
OpenAIgpt-oss-20bopenai/gpt-oss-20b163848192888AMP
OpenAIgpt-oss-120bopenai/gpt-oss-120b163848192161616AMP
DeepSeekDeepSeek-R1-0528deepseek-ai/DeepSeek-R1-0528163848192323232AMP
DeepSeekDeepSeek-R1deepseek-ai/DeepSeek-R1163848192323232AMP
DeepSeekDeepSeek-V3.1deepseek-ai/DeepSeek-V3.1163848192323232AMP
DeepSeekDeepSeek-V3-0324deepseek-ai/DeepSeek-V3-0324163848192323232AMP
DeepSeekDeepSeek-V3deepseek-ai/DeepSeek-V3163848192323232AMP
DeepSeekDeepSeek-V3.1-Basedeepseek-ai/DeepSeek-V3.1-Base163848192323232AMP
DeepSeekDeepSeek-V3-Basedeepseek-ai/DeepSeek-V3-Base163848192323232AMP
DeepSeekDeepSeek-R1-Distill-Llama-70Bdeepseek-ai/DeepSeek-R1-Distill-Llama-70B245768192888bf16
DeepSeekDeepSeek-R1-Distill-Qwen-14Bdeepseek-ai/DeepSeek-R1-Distill-Qwen-14B6553612288888AMP
DeepSeekDeepSeek-R1-Distill-Qwen-1.5Bdeepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B13107216384888AMP
Googlegemma-3-1b-itgoogle/gemma-3-1b-it3276812288888AMP
Googlegemma-3-1b-ptgoogle/gemma-3-1b-pt3276812288888AMP
Googlegemma-3-4b-itgoogle/gemma-3-4b-it13107212288888AMP
Googlegemma-3-4b-ptgoogle/gemma-3-4b-pt13107212288888AMP
Googlegemma-3-12b-itgoogle/gemma-3-12b-it163848192888AMP
Googlegemma-3-12b-ptgoogle/gemma-3-12b-pt655368192888AMP
Googlegemma-3-27b-itgoogle/gemma-3-27b-it491528192888AMP
Googlegemma-3-27b-ptgoogle/gemma-3-27b-pt491528192888AMP
QwenQwen3-0.6BQwen/Qwen3-0.6B3276824576888AMP
QwenQwen3-0.6B-BaseQwen/Qwen3-0.6B-Base3276824576888AMP
QwenQwen3-1.7BQwen/Qwen3-1.7B3276816384888AMP
QwenQwen3-1.7B-BaseQwen/Qwen3-1.7B-Base3276816384888AMP
QwenQwen3-4BQwen/Qwen3-4B3276816384888AMP
QwenQwen3-4B-BaseQwen/Qwen3-4B-Base3276816384888AMP
QwenQwen3-8BQwen/Qwen3-8B3276816384888AMP
QwenQwen3-8B-BaseQwen/Qwen3-8B-Base3276816384888AMP
QwenQwen3-14BQwen/Qwen3-14B3276816384888AMP
QwenQwen3-14B-BaseQwen/Qwen3-14B-Base3276816384888AMP
QwenQwen3-32BQwen/Qwen3-32B245764096888AMP
QwenQwen/Qwen3-30B-A3B-BaseQwen/Qwen3-30B-A3B-Base819281921688AMP
QwenQwen/Qwen3-30B-A3BQwen/Qwen3-30B-A3B819281921688AMP
QwenQwen/Qwen3-30B-A3B-Instruct-2507Qwen/Qwen3-30B-A3B-Instruct-2507819281921688AMP
QwenQwen/Qwen3-235B-A22BQwen/Qwen3-235B-A22B3276816384111AMP
QwenQwen/Qwen3-235B-A22B-Instruct-2507Qwen/Qwen3-235B-A22B-Instruct-25073276816384111AMP
QwenQwen/Qwen3-Coder-30B-A3B-InstructQwen/Qwen3-Coder-30B-A3B-Instruct819281921688AMP
QwenQwen/Qwen3-Coder-480B-A35B-InstructQwen/Qwen3-Coder-480B-A35B-Instruct163848192161616AMP
MetaLlama-3.3-70B-Instruct-Referencemeta-llama/Llama-3.3-70B-Instruct-Reference245768192888bf16
MetaLlama-3.2-3B-Instructmeta-llama/Llama-3.2-3B-Instruct13107224576888AMP
MetaLlama-3.2-3Bmeta-llama/Llama-3.2-3B13107224576888AMP
MetaLlama-3.2-1B-Instructmeta-llama/Llama-3.2-1B-Instruct13107224576888AMP
MetaLlama-3.2-1Bmeta-llama/Llama-3.2-1B13107224576888AMP
MetaMeta-Llama-3.1-8B-Instruct-Referencemeta-llama/Meta-Llama-3.1-8B-Instruct-Reference13107216384888AMP
MetaMeta-Llama-3.1-8B-Referencemeta-llama/Meta-Llama-3.1-8B-Reference13107216384888AMP
MetaMeta-Llama-3.1-70B-Instruct-Referencemeta-llama/Meta-Llama-3.1-70B-Instruct-Reference245768192888bf16
MetaMeta-Llama-3.1-70B-Referencemeta-llama/Meta-Llama-3.1-70B-Reference245768192888bf16
MetaMeta-Llama-3-8B-Instructmeta-llama/Meta-Llama-3-8B-Instruct8192819216168AMP
MetaMeta-Llama-3-8Bmeta-llama/Meta-Llama-3-8B8192819216168AMP
MetaMeta-Llama-3-70B-Instructmeta-llama/Meta-Llama-3-70B-Instruct81928192888bf16
QwenQwen2.5-72B-InstructQwen/Qwen2.5-72B-Instruct245768192888AMP
QwenQwen2.5-72BQwen/Qwen2.5-72B245768192888AMP
QwenQwen2.5-32B-InstructQwen/Qwen2.5-32B-Instruct3276812288888AMP
QwenQwen2.5-32BQwen/Qwen2.5-32B4915212288888AMP
QwenQwen2.5-14B-InstructQwen/Qwen2.5-14B-Instruct3276816384888AMP
QwenQwen2.5-14BQwen/Qwen2.5-14B6553616384888AMP
QwenQwen2.5-7B-InstructQwen/Qwen2.5-7B-Instruct3276816384888AMP
QwenQwen2.5-7BQwen/Qwen2.5-7B13107216384888AMP
QwenQwen2.5-3B-InstructQwen/Qwen2.5-3B-Instruct3276816384888AMP
QwenQwen2.5-3BQwen/Qwen2.5-3B3276816384888AMP
QwenQwen2.5-1.5B-InstructQwen/Qwen2.5-1.5B-Instruct3276816384888AMP
QwenQwen2.5-1.5BQwen/Qwen2.5-1.5B3276816384888AMP
QwenQwen2-72B-InstructQwen/Qwen2-72B-Instruct327688192161616AMP
QwenQwen2-72BQwen/Qwen2-72B327688192161616AMP
QwenQwen2-7B-InstructQwen/Qwen2-7B-Instruct3276816384888AMP
QwenQwen2-7BQwen/Qwen2-7B13107216384888AMP
QwenQwen2-1.5B-InstructQwen/Qwen2-1.5B-Instruct3276816384888AMP
QwenQwen2-1.5BQwen/Qwen2-1.5B13107216384888AMP
Mistral AIMixtral-8x7B-Instruct-v0.1mistralai/Mixtral-8x7B-Instruct-v0.13276832768888bf16
Mistral AIMixtral-8x7B-v0.1mistralai/Mixtral-8x7B-v0.13276832768888bf16
Mistral AIMistral-7B-Instruct-v0.2mistralai/Mistral-7B-Instruct-v0.23276832768888AMP
Mistral AIMistral-7B-v0.1mistralai/Mistral-7B-v0.13276832768888AMP
TekniumOpenHermes-2p5-Mistral-7Bteknium/OpenHermes-2p5-Mistral-7B3276832768888AMP
MetaCodeLlama-7b-hfcodellama/CodeLlama-7b-hf163841638416168AMP
Togetherllama-2-7b-chattogethercomputer/llama-2-7b-chat4096409664648AMP

LoRA Long-context Fine-tuning

OrganizationModel NameModel String for APIContext Length (SFT)Context Length (DPO)Max Batch Size (SFT)Max Batch Size (DPO)Min Batch SizeTraining Precision Type
DeepseekDeepSeek-R1-Distill-Llama-70B-32kdeepseek-ai/DeepSeek-R1-Distill-Llama-70B-32k3276816384111AMP
DeepseekDeepSeek-R1-Distill-Llama-70B-131kdeepseek-ai/DeepSeek-R1-Distill-Llama-70B-131k13107216384111AMP
MetaLlama-3.3-70B-32k-Instruct-Referencemeta-llama/Llama-3.3-70B-32k-Instruct-Reference3276832768111AMP
MetaLlama-3.3-70B-131k-Instruct-Referencemeta-llama/Llama-3.3-70B-131k-Instruct-Reference13107265536111AMP
MetaMeta-Llama-3.1-8B-131k-Instruct-Referencemeta-llama/Meta-Llama-3.1-8B-131k-Instruct-Reference131072131072111AMP
MetaMeta-Llama-3.1-8B-131k-Referencemeta-llama/Meta-Llama-3.1-8B-131k-Reference131072131072111AMP
MetaMeta-Llama-3.1-70B-32k-Instruct-Referencemeta-llama/Meta-Llama-3.1-70B-32k-Instruct-Reference3276832768111AMP
MetaMeta-Llama-3.1-70B-131k-Instruct-Referencemeta-llama/Meta-Llama-3.1-70B-131k-Instruct-Reference13107265536111AMP
MetaMeta-Llama-3.1-70B-32k-Referencemeta-llama/Meta-Llama-3.1-70B-32k-Reference3276832768111AMP
MetaMeta-Llama-3.1-70B-131k-Referencemeta-llama/Meta-Llama-3.1-70B-131k-Reference13107265536111AMP

Full Fine-tuning

OrganizationModel NameModel String for APIContext Length (SFT)Context Length (DPO)Max Batch Size (SFT)Max Batch Size (DPO)Min Batch SizeTraining Precision Type
DeepseekDeepSeek-R1-Distill-Llama-70Bdeepseek-ai/DeepSeek-R1-Distill-Llama-70B245768192161616bf16
DeepseekDeepSeek-R1-Distill-Qwen-14Bdeepseek-ai/DeepSeek-R1-Distill-Qwen-14B6553612288888AMP
DeepseekDeepSeek-R1-Distill-Qwen-1.5Bdeepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B13107216384888AMP
Googlegemma-3-1b-itgoogle/gemma-3-1b-it3276812288888AMP
Googlegemma-3-1b-ptgoogle/gemma-3-1b-pt3276812288888AMP
Googlegemma-3-4b-itgoogle/gemma-3-4b-it13107212288888AMP
Googlegemma-3-4b-ptgoogle/gemma-3-4b-pt13107212288888AMP
Googlegemma-3-12b-itgoogle/gemma-3-12b-it163848192888AMP
Googlegemma-3-12b-ptgoogle/gemma-3-12b-pt655368192888AMP
Googlegemma-3-27b-itgoogle/gemma-3-27b-it491528192161616AMP
Googlegemma-3-27b-ptgoogle/gemma-3-27b-pt491528192161616AMP
QwenQwen3-0.6BQwen/Qwen3-0.6B3276824576888AMP
QwenQwen3-0.6B-BaseQwen/Qwen3-0.6B-Base3276824576888AMP
QwenQwen3-1.7BQwen/Qwen3-1.7B3276816384888AMP
QwenQwen3-1.7B-BaseQwen/Qwen3-1.7B-Base3276816384888AMP
QwenQwen3-4BQwen/Qwen3-4B3276816384888AMP
QwenQwen3-4B-BaseQwen/Qwen3-4B-Base3276816384888AMP
QwenQwen3-8BQwen/Qwen3-8B3276816384888AMP
QwenQwen3-8B-BaseQwen/Qwen3-8B-Base3276816384888AMP
QwenQwen3-14BQwen/Qwen3-14B3276816384888AMP
QwenQwen3-14B-BaseQwen/Qwen3-14B-Base3276816384888AMP
QwenQwen3-32BQwen/Qwen3-32B245764096161616AMP
QwenQwen/Qwen3-30B-A3B-BaseQwen/Qwen3-30B-A3B-Base81928192888AMP
QwenQwen/Qwen3-30B-A3BQwen/Qwen3-30B-A3B81928192888AMP
QwenQwen/Qwen3-30B-A3B-Instruct-2507Qwen/Qwen3-30B-A3B-Instruct-250781928192888AMP
QwenQwen/Qwen3-Coder-30B-A3B-InstructQwen/Qwen3-Coder-30B-A3B-Instruct81928192888AMP
MetaLlama-3.3-70B-Instruct-Referencemeta-llama/Llama-3.3-70B-Instruct-Reference245768192161616bf16
MetaLlama-3.2-3B-Instructmeta-llama/Llama-3.2-3B-Instruct13107224576888AMP
MetaLlama-3.2-3Bmeta-llama/Llama-3.2-3B13107224576888AMP
MetaLlama-3.2-1B-Instructmeta-llama/Llama-3.2-1B-Instruct13107224576888AMP
MetaLlama-3.2-1Bmeta-llama/Llama-3.2-1B13107224576888AMP
MetaMeta-Llama-3.1-8B-Instruct-Referencemeta-llama/Meta-Llama-3.1-8B-Instruct-Reference13107216384888AMP
MetaMeta-Llama-3.1-8B-Referencemeta-llama/Meta-Llama-3.1-8B-Reference13107216384888AMP
MetaMeta-Llama-3.1-70B-Instruct-Referencemeta-llama/Meta-Llama-3.1-70B-Instruct-Reference245768192161616bf16
MetaMeta-Llama-3.1-70B-Referencemeta-llama/Meta-Llama-3.1-70B-Reference245768192161616bf16
MetaMeta-Llama-3-8B-Instructmeta-llama/Meta-Llama-3-8B-Instruct8192819216168AMP
MetaMeta-Llama-3-8Bmeta-llama/Meta-Llama-3-8B8192819216168AMP
MetaMeta-Llama-3-70B-Instructmeta-llama/Meta-Llama-3-70B-Instruct81928192161616bf16
QwenQwen2-7B-InstructQwen/Qwen2-7B-Instruct3276816384888AMP
QwenQwen2-7BQwen/Qwen2-7B13107216384888AMP
QwenQwen2-1.5B-InstructQwen/Qwen2-1.5B-Instruct3276816384888AMP
QwenQwen2-1.5BQwen/Qwen2-1.5B13107216384888AMP
Mistral AIMixtral-8x7B-Instruct-v0.1mistralai/Mixtral-8x7B-Instruct-v0.13276832768161616bf16
Mistral AIMixtral-8x7B-v0.1mistralai/Mixtral-8x7B-v0.13276832768161616bf16
Mistral AIMistral-7B-Instruct-v0.2mistralai/Mistral-7B-Instruct-v0.23276832768888AMP
Mistral AIMistral-7B-v0.1mistralai/Mistral-7B-v0.13276832768888AMP
TekniumOpenHermes-2p5-Mistral-7Bteknium/OpenHermes-2p5-Mistral-7B3276832768888AMP
MetaCodeLlama-7b-hfcodellama/CodeLlama-7b-hf163841638416168AMP
Togetherllama-2-7b-chattogethercomputer/llama-2-7b-chat4096409664648AMP