GraySoft
Download guIDE →
Projects Models Compare Cloud benchmarks FAQ Download guIDE →

DeepSeek cloud models

35 models tracked via Artificial Analysis. Compare cloud performance, then find local GGUF versions in the GraySoft model catalog.

ModelIntelligenceSpeed (tok/s)
DeepSeek V4.1 Flash (Reasoning, Max Effort)39.5267.06
DeepSeek V4 Pro 0813 (Reasoning, Max Effort)36.369.678
DeepSeek V4 Flash Vision (Reasoning, Max Effort)35216.297
DeepSeek V4 Flash 0731 (Reasoning, Max Effort)34.5239.381
DeepSeek V4 Pro (Reasoning, Max Effort)30.966.356
DeepSeek V4 Pro (Reasoning, High Effort)30.168.431
DeepSeek V4 Flash (Reasoning, High Effort)24.80
DeepSeek V4 Flash (Reasoning, Max Effort)24.60
DeepSeek V3.2 (Reasoning)21.50
DeepSeek V4 Pro (Non-reasoning)20.866.05
DeepSeek V4 Flash (Non-reasoning)18.90
DeepSeek V3.2 Exp (Reasoning)16.60
DeepSeek V3.2 (Non-reasoning)160
DeepSeek V3.1 Terminus (Reasoning)15.40
DeepSeek V3.2 Speciale14.50
DeepSeek V3.2 Exp (Non-reasoning)13.90
DeepSeek V3.1 Terminus (Non-reasoning)13.90
DeepSeek V3.1 (Non-reasoning)13.70
DeepSeek V3.1 (Reasoning)13.50
DeepSeek R1 0528 (May '25)13.10
DeepSeek R1 (Jan '25)11.40
DeepSeek V3 03249.70
DeepSeek V3 (Dec '24)8.50
DeepSeek R1 Distill Qwen 32B8.40
DeepSeek R1 0528 Qwen3 8B8.10
DeepSeek R1 Distill Llama 70B7.90
DeepSeek R1 Distill Qwen 14B7.80
DeepSeek-V2.5 (Dec '24)6.60
DeepSeek-V2.56.60
DeepSeek R1 Distill Llama 8B6.50
DeepSeek-Coder-V260
DeepSeek R1 Distill Qwen 1.5B5.50
DeepSeek-V2-Chat5.50
DeepSeek LLM 67B Chat (V1)5.30
DeepSeek Coder V2 Lite Instruct5.30

Run models locally with guIDE

Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.

Download guIDE → · Browse 524k+ models · Compare models