214 Downloads Updated 20 hours ago
DeepSeek-V4.1-Flash is an advanced tool designed to enhance search capabilities, providing users with faster and more accurate results.
ollama run deepseek-v4.1-flash:cloud
Applications
ollama launch claude --model deepseek-v4.1-flash:cloud
ollama launch opencode --model deepseek-v4.1-flash:cloud
ollama launch hermes --model deepseek-v4.1-flash:cloud
ollama launch openclaw --model deepseek-v4.1-flash:cloud
Models
View all →Readme
DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression
DeepSeek-V4.1-Flash is an advanced tool designed to enhance search capabilities, providing users with faster and more accurate results. It’s a multimodal Mixture-of-Experts (MoE) model with 552B backbone parameters and support for contexts of up to one million tokens.
DeepSeek-V4.1-Flash uses a Causal Encoder-Decoder (CED) design, stacking a 20-layer causal encoder on top of a 20-layer decoder for 40 Transformer layers total. Under CED, the decoder’s global KV cache is built by projecting the encoder’s final hidden states, instead of being computed from each decoder layer’s hidden states as it normally would be. As a result, only 8B parameters are active per token during prefill and 16B during decode, which cuts costs significantly for agentic workloads that process a lot of input.

DeepSeek-V4.1-Flash builds on Compressed Sparse Attention 2 (CSA2), where every attention layer runs in one of three fixed modes - Full, Reindex, or Reuse - letting layers share main KV, share indexer K, and recycle Top-K sparse-attention indices. In the decoder, a Hierarchical Sparse Indexer caps later indexing layers to a candidate pool built by the first Full Mode layer, so indexer cost at deeper layers stays bounded no matter how long the context gets. Together with FP4 main KV caching (E2M1 values, one E4M3 scale per 16 channels), this brings the global KV cache down to 890 bytes per token - about a quarter of DeepSeek-V4-Flash.
| Benchmark (Metric) | Opus-5.0 | GPT-5.6 Sol | K3 | GLM-5.3 | DS-V4-Pro | DS-V4-Flash | DS-V4.1-Flash |
|---|---|---|---|---|---|---|---|
| Reasoning | |||||||
| GPQA Diamond (Pass@1) | 93.4 | 94.1 | 92.9 | 88.1 | 92.4 | 89.9 | 90.9 |
| HLE (Pass@1) | 56.3 | 44.5 | 43.5 | 42.0† | 42.7† | 37.8† | 36.8 (39.1†) |
| Codeforces (Rating) | — | — | — | — | 3348 | 3289 | 3471 |
| MathArena Apex (Pass@1) | — | — | 65.6 | — | 65.3 | 58.6 | 65.6 |
| Agentic | |||||||
| Terminal-Bench 2.1 (Pass@1) | 89.1 | 88.8 | 88.3 | 88.2 | 87.9 | 82.7 | 90.6 |
| Terminal-Bench 3.0 (Pass@1) | 43.3 | 34.4 | 17.7 | 28.3 | 11.8 | 7.6 | 30.0 |
| Terminal-Bench 4.0 (Pass@1) | 51.8 | 39.9 | 12.6 | 37.9 | 12.4 | 7.0 | 31.2 |
| DeepSWE v1.1 (Resolved) | 74.0 | 73.0 | 67.5 | 66.9 | 62.7 | 54.4 | 74.2 |
| ProgramBench (Almost@1) | 37.0 | 23.0 | 17.5 | 19.0 | 15.5 | — | 20.3 |
| NL2Repo-Bench (Score) | 75.3 | 56.8 | 58.0 | 58.0 | 61.5 | 54.2 | 64.0 |
| CyberGym (Pass@1) | — | 84.5 | 80.0 | 84.5 | 83.3 | 76.7 | 88.1 |
| SEC-Bench Pro (Pass@1) | — | 74.3 | — | — | 56.4 | 30.9 | 62.8 |
| ExploitGym (Pass@1) | 22.1 | 33.7 | — | 15.0 | 5.4 | 1.8 | 15.3 |
| HLE w/ tools (Pass@1) | 63.6 | — | 59.8 | 62.5 | 60.0 | 51.5 | 63.9 |
| AutomationBench (Pass@1) | 50.3 | 45.8 | 46.7 | 48.8 | 43.2 | 37.7 | 54.8 |
| Agent’s Last Exam (Pass@1) | 28.6 | 26.7 | 27.6 | 28.5 | 25.7 | 25.2 | 31.8 |
| Chartography w/ tools (Pass@1) | 84.0 | 79.9 | 68.1 | — | — | — | 78.9 |
| BabyVision w/ tools (Pass@1) | 94.1 | 88.9 | 85.7 | — | — | — | 89.6 |
| ZeroBench-main w/ tools (Pass@5) | 52.0 | 53.0 | 41.0 | — | — | — | 49.0 |