Models - Sail Research

Sail serves the reference model by default. Any quantization is called out explicitly, and we always link to the exact weights we’re using.

Core models

Model	Model ID	Image	LoRA	Reasoning
Kimi K3 Moonshot AI	`moonshotai/Kimi-K3` ReferenceView checkpoint
GLM-5.3 Z.ai	`zai-org/GLM-5.3` ReferenceView checkpoint
GLM-5.3-Flash Z.ai	`zai-org/GLM-5.3-Flash` NVIDIA quantView checkpoint
DeepSeek V4.1 Flash 1017 DeepSeek	`deepseek-ai/DeepSeek-V4.1-Flash-1017` ReferenceView checkpoint
DeepSeek V4 Pro 0813 DeepSeek	`deepseek-ai/DeepSeek-V4-Pro-0813` ReferenceView checkpoint
DeepSeek V4 Flash 0731 DeepSeek	`deepseek-ai/DeepSeek-V4-Flash-0731` ReferenceView checkpoint
Gemma 4 31B IT Google	`google/gemma-4-31B-it` ReferenceView checkpoint
Gemma 4 31B IT (NVFP4) Google	`nvidia/Gemma-4-31B-IT-NVFP4` NVIDIA quantView checkpoint
Gemma 4 12B IT Google	`google/gemma-4-12B-it` ReferenceView checkpoint

Flex-only models

Models served with the flex completion window exclusively.

Model	Model ID	Image	LoRA	Reasoning
Qwen3.6 35B A3B Qwen	`Qwen/Qwen3.6-35B-A3B` ReferenceView checkpoint

Notes

If we offer multiple quantizations of a model, we list them as separate model IDs.
Use GET /v1/models to confirm runtime availability for your API key.
For per-model rates by completion window, see Pricing.