DeepSeekopen-weight
Released: 2026-10-17Artificial AnalysisDeepSeek V4.1 Flash 1017
OpenRouter ranked #1 foundation model processing 58.04T tokens weekly. Deployed by deepseek.
Intelligence Index
32.6/ 100
Artificial Analysis Intelligence Index (independent composite)Throughput Speed
171tok/sec
Output streaming throughputTime to First Answer
45s
Time to first answer token, including any thinkingCost to Run the Index
0.60USD
Artificial Analysis cost of the full Index runMeasured results · Artificial Analysis
Independent evaluations of DeepSeek V4.1 Flash 1017 (Max configuration). Values are reported as published; nothing here is estimated or derived. Snapshot 2026-11-15.
- Intelligence Index
- 32.6
- Cost to run the Index
- $0.60
- Time per task
- 13.5 min
- Omniscience index
- 0.9
- Reasoning model
- Yes
- Cached-input price
- $0.006 / 1M
Reasoning & knowledge
- AA-Omniscience Accuracy39.6%
- CritPt24.0%
- CritPt v2 (Hard)25.0%
- HLE-Diamond25.3%
- Humanity's Last Exam43.3%
Coding & terminal agents
- FrontierCode 1.143.7%
- FrontierCode-AA 2.028.5%
- SciCode55.9%
- Terminal-Bench 4.033.8%
- Terminal-Bench-Science 0.112.9%
Agentic work
- AA-Briefcase v1.1 (Elo)1570 Elo
- AutomationBench-AA66.6%
- GDPval-AA (normalized Elo)1746 Elo
- Multimodal Knowledge Agent (MKA)38.5%
Long context & documents
Health & safety
Reasoning-effort variants evaluated
| Variant | Intelligence Index | Cost to run Index | Output speed (tok/s) |
|---|---|---|---|
| Max | 32.6 | $0.60 | 171.1 |
| High | 28.6 | $0.46 | 171.8 |
| Non-reasoning | 18.5 | $0.43 | 149.8 |
Headline Evaluation Breakdown
Six independently measured evaluations (Artificial Analysis)Reasoning · CritPt v2 (Hard)25.0%
Coding & agents · Terminal-Bench 4.033.8%
Long context · AA-LCR82.7%
Knowledge · Humanity's Last Exam43.3%
Sourced Benchmark Evaluations (18)
Verified performance across authoritative benchmarks with provenance tracking.| Benchmark | Status | Score | Trust | Date | Source Type | Provenance |
|---|---|---|---|---|---|---|
| HLE-Diamond | Active | 25.3% | 93 | 2026-11-15 | independent | Source ↗ |
| VIALS-AA | Active | 40.4% | 93 | 2026-11-15 | independent | Source ↗ |
| AutomationBench-AA | Active | 66.6% | 93 | 2026-11-15 | independent | Source ↗ |
| Multimodal Knowledge Agent (MKA) | Active | 38.5% | 93 | 2026-11-15 | independent | Source ↗ |
| MLCR-AA (Medical Long Context Reasoning) | Active | 26.7% | 93 | 2026-11-15 | independent | Source ↗ |
| GDP.pdf | Active | 17.2% | 93 | 2026-11-15 | independent | Source ↗ |
| FrontierCode-AA 2.0 | Active | 28.5% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-Omniscience (Non-Hallucination) | Active | 35.9% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-Omniscience (Accuracy) | Active | 39.6% | 93 | 2026-11-15 | independent | Source ↗ |
| FrontierCode | Active | 43.7% | 98 | 2026-11-15 | independent | Source ↗ |
| CritPt v2 (Hard) | Active | 25% | 93 | 2026-11-15 | independent | Source ↗ |
| CritPt | Active | 24% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-LCR (Long Context Reasoning) | Active | 82.7% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench-Science 0.1 | Active | 12.9% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench 4.0 | Active | 33.8% | 93 | 2026-11-15 | independent | Source ↗ |
| SciCode | Active | 55.9% | 93 | 2026-11-15 | independent | Source ↗ |
| MMMU-Pro | Active | 80.6% | 93 | 2026-11-15 | independent | Source ↗ |
| Humanity's Last Exam | Active | 43.3% | 93 | 2026-11-15 | independent | Source ↗ |
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run deepseek-v4-1-flash-1017
# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-v4-1-flash-1017",
"messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
}'Evaluation & Sourcing Notes
Scores listed for DeepSeek V4.1 Flash 1017 represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.
All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.