Introduction
DeepSeek-V4.1-Flash-1017 is a insignificant update to DeepSeek-V4.1-Flash, delivering significant improvements throughout reasoning, coding, agentic tasks, and ocular benchmarks. Alongside these gains, the example reduces project implementation period and output length, enabling users to complete tasks faster and at lesser disbursal and advancing the performance–cost Pareto frontier throughout multiple delegate harnesses.
| Benchmark | DS-V4.1-Flash-1017 | DS-V4.1-Flash | K3 | GLM-5.3 | GPT-5.6 Sol | Opus 5 |
|---|---|---|---|---|---|---|
| HLE w/o tools | 42.6 (44.0†) | 36.8 (39.1†) | 43.5 | 42.0† | 44.5 | 56.3 |
| Terminal-Bench 4.0 | 42.3 | 31.2 | 12.6 | 37.9 | 39.9 | 51.8 |
| DeepSWE v1.1 | 77.0 | 74.2 | 67.5 | 66.9 | 73.0 | 74.0 |
| ProgramBench | 29.1 | 20.3 | 17.5 | 19.0 | 23.0 | 37.0 |
| NL2Repo-Bench | 68.2 | 65.4 | 58.0 | 58.0 | 56.8 | 75.3 |
| Automation-Bench v1.0.6 | 60.5 | 54.8 | 46.7 | 48.8 | 45.8 | 50.3 |
| Agents’ Last Exam | 35.4 | 31.8 | 27.6 | 28.5 | 26.7 | 28.6 |
| Apex Bench v1.1 | 72.1 | 64.6 | 47.8 | 56.6 | 54.3 | 65.8 |
| Chartography (w/tools) | 83.0 | 78.9 | 68.1 | — | 79.9 | 84.0 |
| BabyVision (w/tools) | 92.5 | 89.6 | 85.7 | — | 88.9 | 94.1 |
| ZeroBench-main (w/tools) | 51.0 | 49.0 | 41.0 | — | 53.0 | 52.0 |
† denotes the text-only subset of HLE. We use the Claude Code harness for Terminal-Bench 4.0 and the Mini-SWE-V2 harness for DeepSWE v1.1.
Prompt Encoding
This publish does not contain a Jinja-format conversation template. The encoding directory contains a self-contained Python citation implementation (encoding.py) alongside test cases for multi-turn conversations, tool calling, thinking mode, numeric reasoning effort, mid-conversation scheme messages, and interleaved depiction content.
For manufacturing use, we additionally publish deepseek-recipe, a set of Rust libraries alongside Python bindings that provides the identical immediate format as a maintained, protocol-aware toolkit. It converts Messages, Chat Completions, and Responses API requests into the Conversation format, encodes them into DeepSeek V4 and V4.1 prompts or token IDs, and parses example output rear into complete or streamed responses — covering thinking, tool calls, images, and generation settings. Model inference, tool execution, and HTTP transport are remaining to the caller.
Minimal Inference
Please mention to the inference directory for instructions on importance conversion and operating conclusion locally.
Recommended sampling parameters:
| Parameter | Value |
|---|---|
| temperature | 1.0 |
| top_p | 1.0 |
| context_window | 1M tokens |
| max_tokens | ≥ 256K |
License
This repository and the example weights are licensed under the MIT License.
Citation
@misc{deepseekai2026deepseekv41flash, title={DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression}, author={DeepSeek-AI}, year={2026}, } Contact
If you have any questions, delight lift an matter or communication us at [email protected].
