DeepSeek V4.1 Flash 1017 is DeepSeek's open-weight AI model, released 17/10/2026. DeepSeek's model card compares the 1017 weights update with the September V4.1-Flash, Kimi K3, GLM-5.3, GPT-5.6 Sol and Claude Opus 5. The 1017 build leads on DeepSWE v1.1 (77.0), Automation-Bench and Apex Bench, and sits well below Claude Opus 5 on Terminal-Bench 4.0, HLE and ProgramBench. The card does not group the rows into categories, so the categories here are ours.
Visit DeepSeek V4.1 Flash 1017AI-Powered
Leverages advanced AI technology to deliver cutting-edge capabilities and results.
Fast & Efficient
Optimized performance ensures quick results without compromising on quality.
Purpose-Built
Specifically designed for llms tasks and workflows.
How does DeepSeek V4.1 Flash 1017 score on independent benchmarks?
DeepSeek V4.1 Flash 1017 ranks #18 of 28 models on the Artificial Analysis Intelligence Index (v5.0.1) with a score of 32.6, as measured on 22 November 2026.
Where does DeepSeek V4.1 Flash 1017 rank against other models?
How good is DeepSeek V4.1 Flash 1017 at agentic coding?
How does DeepSeek V4.1 Flash 1017 do on individual tests?
| Evaluation (run by Artificial Analysis) | DeepSeek V4.1 Flash 1017 | Best of any model |
|---|---|---|
| Terminal-Bench 4.0 | 33.8% | 67.2% |
| FrontierCode 2.0 | 28.5% | 48.8% |
| Humanity's Last Exam | 43.3% | 74.3% |
| GDPval-AA (Elo) | 1746 | 1866 |
How has DeepSeek V4.1 Flash 1017's score changed over time?
We have tracked DeepSeek V4.1 Flash 1017 on Intelligence Index v5.0.1 since 22/11/2026. A point is added each day the score, coding index or price changes, so this chart fills in over time.
What does the DeepSeek V4.1 Flash 1017 system card report?
DeepSeek's model card compares the 1017 weights update with the September V4.1-Flash, Kimi K3, GLM-5.3, GPT-5.6 Sol and Claude Opus 5. The 1017 build leads on DeepSWE v1.1 (77.0), Automation-Bench and Apex Bench, and sits well below Claude Opus 5 on Terminal-Bench 4.0, HLE and ProgramBench. The card does not group the rows into categories, so the categories here are ours.
These are DeepSeek's own published results (self-reported, with the vendor's harness and settings), so they are not directly comparable with other labs' numbers. Source: DeepSeek-V4.1-Flash-1017 (Hugging Face model card) (17/10/2026). Last checked 22/11/2026.
Reasoning & maths
| Benchmark | DeepSeek V4.1 Flash 1017 | DeepSeek-V4.1-Flash (September) | Kimi K3 | GLM-5.3 |
|---|---|---|---|---|
| HLE (no tools)Full set; text-only subset score is 44.0. GLM-5.3 figure is the text-only subset | 42.6% | 36.8% | 43.5% | 42% |
Coding
| Benchmark | DeepSeek V4.1 Flash 1017 | DeepSeek-V4.1-Flash (September) | Kimi K3 | GLM-5.3 |
|---|---|---|---|---|
| Terminal-Bench 4.0Claude Code harness | 42.3% | 31.2% | 12.6% | 37.9% |
| DeepSWE v1.1Mini-SWE-V2 harness | 77% | 74.2% | 67.5% | 66.9% |
| ProgramBench | 29.1% | 20.3% | 17.5% | 19% |
| NL2Repo-Bench | 68.2% | 65.4% | 58% | 58% |
Agents & tool use
| Benchmark | DeepSeek V4.1 Flash 1017 | DeepSeek-V4.1-Flash (September) | Kimi K3 | GLM-5.3 |
|---|---|---|---|---|
| Automation-Bench v1.0.6 | 60.5% | 54.8% | 46.7% | 48.8% |
| Agents' Last Exam | 35.4% | 31.8% | 27.6% | 28.5% |
| Apex Bench v1.1 | 72.1% | 64.6% | 47.8% | 56.6% |
Multimodal
| Benchmark | DeepSeek V4.1 Flash 1017 | DeepSeek-V4.1-Flash (September) | Kimi K3 | GLM-5.3 |
|---|---|---|---|---|
| Chartography (w/ tools)With tools | 83% | 78.9% | 68.1% | n/a |
| BabyVision (w/ tools)With tools | 92.5% | 89.6% | 85.7% | n/a |
| ZeroBench-main (w/ tools)With tools | 51% | 49% | 41% | n/a |
What did safety testing find for DeepSeek V4.1 Flash 1017?
Safety level: No vendor safety rating on the model card
- The model card page read lists benchmark results, recommended sampling (temperature 1.0, top_p 1.0) and an MIT licence; no safety evaluation was found on it.
Independent figures: Artificial Analysis, refreshed daily on our AI model benchmarks page. Full analysis: our DeepSeek V4.1 Flash 1017 deep dive.
DeepSeek Model Timeline
1M tokens (max tokens listed as at least 256K) context
1M tokens context
Specifications
Related Tools
Claude Opus 5.5
Claude Opus 5.5, launched 22 September 2026, is Anthropic's current Opus-line flagship, replacing Opus 5. It performs at roughly the level of Claude Fable 5.1 on most work while costing 40% less to run than Opus 5, at $4 input / $20 output per million tokens (both 20% down, cache reads cut 60% to $0.20). It leads on Terminal-Bench 4.0, GDPval-AA v2.1 and Humanity's Last Exam, and is the first Opus model to ship with Fable-5.1-class cybersecurity, biology and distillation safeguards.
Claude Sonnet 5.5
Claude Sonnet 5.5, released 28 September 2026, is the second model in Anthropic's Claude 5.5 family and the faster, lower-cost complement to Claude Opus 5.5. It keeps Sonnet 5's $2 input / $10 output per million tokens ($0.20 cache reads) while generating output over 30% faster, and scores 70.6% on Terminal-Bench 4.0, above Opus 5.5's 66.4%. It has a 1M-token context window, 128K max output, adaptive thinking, and is the first Sonnet with cyber and anti-distillation classifiers.
GPT-6.1 Sol
GPT-6.1 Sol, launched at OpenAI DevDay on 29 September 2026, is OpenAI's mid-tier workhorse model. OpenAI says it delivers nearly the same intelligence as GPT-6 Astra at one-fifth of Astra's token price: $2 input / $10 output per million tokens with cached input cut to $0.10. It scores 75.2% on DeepSWE v1.1 (high effort) and 71.4% on OSWorld 2.0, has a 1.05M-token context window and 128K max output, and is available in the API plus ChatGPT Work and Codex.
