# DeepSeek V4.1 Flash 1017

DeepSeek · Open Weight · rank 23 · bench-align-v5

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelchoice.dev/models/deepseek-v4-1-flash  
JSON: https://modelchoice.dev/api/model/deepseek-v4-1-flash

## Facts

| Field | Value | Evidence |
| --- | --- | --- |
| Canonical id | `deepseek-v4-1-flash` | — |
| Overall score | 70 | Observed 2026-11-21 · source benchlm:models |
| Verified score (BenchLM): a separate BenchLM figure, not the overall score | 75 | Observed 2026-11-21 · source benchlm:models |
| Context window | 1M tokens | Observed 2026-11-21 · source benchlm:models |
| Release date | 2026-10-17 | Observed 2026-11-21 · source benchlm:models |
| Access type | Open Weight | — |
| Blended $/1M (75% input / 25% output) | $0.52 | Derived from the input and output rates below |

## Capability evidence

Seven axes from the ranking source. An axis the source did not score is unavailable, not zero. Beside each score is BenchLM's separate verified score for that axis; the two are published side by side and are not interchangeable.

| Axis | Score | Evidence | Verified (BenchLM) | Verified evidence |
| --- | --- | --- | --- | --- |
| Agentic | 75.4 | Observed 2026-11-21 · source benchlm:models | 81.1 | Observed 2026-11-21 · source benchlm:models |
| Coding | 65 | Observed 2026-11-21 · source benchlm:models | Unavailable | Unavailable · source benchlm:models |
| Knowledge | 66.7 | Observed 2026-11-21 · source benchlm:models | 66.7 | Observed 2026-11-21 · source benchlm:models |
| Reasoning | 81.6 | Observed 2026-11-21 · source benchlm:models | 81.6 | Observed 2026-11-21 · source benchlm:models |
| Multimodal & Grounded | 77.3 | Observed 2026-11-21 · source benchlm:models | 77.3 | Observed 2026-11-21 · source benchlm:models |
| Instruction Following | Unavailable | Unavailable · source benchlm:models | Unavailable | Unavailable · source benchlm:models |
| Math | 64.4 | Observed 2026-11-21 · source benchlm:models | 64.4 | Observed 2026-11-21 · source benchlm:models |

## Runtime service evidence

Measured values with the date they were observed. Nothing is inferred from a sibling model or a provider claim. Regional or per-endpoint measurements appear only when the API supplies them; none are modelled.

| Measurement | Value | Observed | Last good | Evidence |
| --- | --- | --- | --- | --- |
| Time to first token | 44.28 s | 2026-11-21 | 2026-11-21 | Observed 2026-11-21 · source benchlm:speed |
| Throughput | 163 tok/s | 2026-11-21 | 2026-11-21 | Observed 2026-11-21 · source benchlm:speed |

## Endpoint and price matrix

Every published price component, including cache reads and writes.

| Component | USD | Evidence |
| --- | --- | --- |
| Input / 1M tokens | $0.30 | Observed 2026-11-21 · source benchlm:pricing |
| Output / 1M tokens | $1.20 | Observed 2026-11-21 · source benchlm:pricing |
| Cache read / 1M tokens | $0.006 | Observed 2026-11-21 · source benchlm:pricing |
| Cache write / 1M tokens | Unavailable | Unavailable · source openrouter:pricing |
| Blended / 1M (75% input / 25% output) | $0.52 | Derived — from the input and output rates above; it has no source record of its own |
| Cost per successful task (LiveBench) | Unavailable | Unavailable · source livebench:table |

**Pricing note.** DeepSeek's pricing page read October 17, 2026 lists `deepseek-flash` with model version DeepSeek-V4.1-Flash-1017 at the same peak and off-peak rates, and DeepSeek's October 17, 2026 changelog says the upgrade leaves the API calling method and pricing unchanged (https://api-docs.deepseek.com/updates). DeepSeek's pricing page checked September 10, 2026 lists `deepseek-flash` (model version DeepSeek-V4.1-Flash) with peak and off-peak rates, effective 04:00 UTC on September 10, 2026. The stored values are the peak rates: $0.30 cache-miss input / $0.006 cache-hit input / $1.20 output per million tokens. Off-peak rates are half of peak: $0.15 / $0.003 / $0.60. Peak hours are 01:00-04:00 and 06:00-10:00 UTC, Monday through Friday; all other hours are off-peak. The page lists a 1M context length, a 384K maximum output, and a 2,500-request concurrency limit. The legacy `deepseek-v4-flash` and `deepseek-v4-flash-vision-exp` IDs route to this model at the same Flash price, and DeepSeek's September 10, 2026 announcement said `deepseek-v4-pro` requests would join them from 04:00 UTC on September 14, 2026. DeepSeek's pricing page, read September 17, 2026, says V4 Pro API service continues after September 14, 2026 with the billing method unchanged, and that DeepSeek will give notice of any change; the September 10 post had not been edited when read the same day (https://api-docs.deepseek.com/news/news260910, https://api-docs.deepseek.com/quick_start/pricing). MIT-licensed weights are also published on Hugging Face for self-hosting.

## Workload-aware monthly cost example

10 conversations per day × 8 messages × 22 active days, 1200 input and 400 output tokens per message, no cache. Derived here from the published rates above by this site's own calculator — not a figure any source published.

| Field | Value |
| --- | --- |
| Modelled monthly cost | $1.48 |
| Modelled tokens | 2.82M |

## Benchmark record

75 matched benchmark rows with their published value, unit, and provenance.

### Knowledge

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| GPQA (Graduate-Level Google-Proof Q&A) | 90.9 | 448 questions | Graduate level | [GPQA: A Graduate-Level Google-Proof Q&A Benchmark](https://arxiv.org/abs/2311.12022) |
| GPQA-D (GPQA Diamond) | 90.9 | Graduate-level science questions | Graduate level | [GPQA: A Graduate-Level Google-Proof Q&A Benchmark](https://arxiv.org/abs/2311.12022) |
| HLE (Humanity's Last Exam) | 42.6 | Expert-level questions | Frontier expert level | [Humanity's Last Exam](https://lastexam.ai/) |
| Artificial Analysis Intelligence Index | 39.5 | Cross-benchmark intelligence index | Display-only external reference | [Artificial Analysis](https://artificialanalysis.ai/) |
| AA-HLE (Artificial Analysis Humanity's Last Exam) | 39.2 | Expert-level questions | Frontier expert reasoning | [Artificial Analysis Humanity's Last Exam Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/hle) |
| AA-Omniscience Index (Artificial Analysis Omniscience Index) | -5.3 | Knowledge questions | Broad factual knowledge | [AA-Omniscience: Knowledge and Hallucination Benchmark](https://artificialanalysis.ai/evaluations/omniscience) |
| AA-Omniscience Accuracy (Artificial Analysis Omniscience Accuracy) | 46.4 | Knowledge questions | Broad knowledge | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/grok-4-3) |
| AA-Omniscience Hallucination Rate (Artificial Analysis Omniscience Hallucination Rate) | 96.5 | Knowledge questions | Factuality | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/grok-4-3) |

### Coding

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| Terminal-Bench 2.1 | 90.6 | Terminal-based software-agent tasks | Professional software engineering | [Terminal-Bench](https://www.tbench.ai/) |
| Codeforces (Codeforces Rating) | 3471.0 | Competitive programming contests | Elite competitive programming | [DeepSeek-V4 Technical Report](https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro/blob/main/DeepSeek_V4.pdf) |
| OpenHarmony Bench v1.0 | 60.3 | 153 app-development and bug-fix tasks | End-to-end OpenHarmony application development | [OpenHarmony Bench: Evaluating LLMs and Coding Agents on OpenHarmony App Development](https://arxiv.org/abs/2608.16022) |
| ProgramBench (ProgramBench: Can Language Models Rebuild Programs From Scratch?) | 29.1 | 200 program reconstruction tasks | Full-repository software architecture | [ProgramBench: Can Language Models Rebuild Programs From Scratch?](https://programbench.com/static/paper.pdf) |
| Bug Hunt Bench | 39.1 | 105 planted bugs across two production TypeScript repositories | Blind production-repository bug finding and repair | [Bug Hunt Bench method, caveats, and definitions](https://bughunt.productcompass.pm/method) |
| NL2Repo | 68.2 | Natural language to repository tasks | System-level software comprehension | [MiniMax M2.7: Early Echoes of Self-Evolution](https://www.minimax.io/news/minimax-m27-en) |
| AA-SciCode (Artificial Analysis SciCode) | 51.9 | Scientific coding subproblems | Scientific programming | [Artificial Analysis SciCode Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/scicode) |
| DeepSWE | 77.0 | 113 software engineering tasks across 91 repositories and 5 languages | Long-horizon software engineering | [DeepSWE benchmark blog](https://deepswe.datacurve.ai/blog) |
| VulcanBench Frontier v4 | 88.7 | 23 behavioural-reconstruction coding tasks | Professional multi-file software engineering | [VulcanBench Frontier v4 leaderboard](https://vulcanbench.com/leaderboard.html) |
| OpenHarmony Bench v1.2 | 71.28 | App-development, bug-fix and migration tasks (count not published) | End-to-end OpenHarmony application development and migration | [OpenHarmony Bench: Evaluating LLMs and Coding Agents on OpenHarmony App Development](https://arxiv.org/abs/2608.16022) |
| FrontierCode 2.0 | 20.2 | 132 private software-engineering tasks | Frontier coding-agent quality | [FrontierCode 2.0](https://cognition.com/blog/frontier-code-2.0) |
| ScarfBench (ScarfBench: Cross-Framework Application Migration in Enterprise Java) | 69.0 | 204 framework migration tasks across 34 enterprise Java applications | Whole-application framework migration | [ScarfBench: A Benchmark for Cross-Framework Application Migration in Enterprise Java](https://arxiv.org/abs/2605.06754) |

### Mathematics

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| Apex | 65.6 | Advanced mathematical reasoning | Frontier math reasoning | [DeepSeek-V4 Technical Report](https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro/blob/main/DeepSeek_V4.pdf) |
| FrontierMath v2 (Tiers 1-3) (FrontierMath v2 Tiers 1-3) | 67.368 | 285 private advanced mathematics problems | From olympiad-plus to early research mathematics | [FrontierMath v2 benchmark hub](https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2) |
| FrontierMath v2 (Tier 4) (FrontierMath v2 Tier 4) | 26.829 | 41 private extreme-difficulty mathematics problems | Research-level mathematics requiring hours or days of expert work | [FrontierMath Tier 4 v2 leaderboard](https://epoch.ai/benchmarks/frontiermath-tier-4-v2?view=graph&tab=leaderboard) |

### Reasoning

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| AA-LCR (Artificial Analysis Long Context Reasoning) | 84.0 | Long-context reasoning tasks | Long-context reasoning | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/grok-4-3) |
| MLCR-AA (Medical Long Context Reasoning (MLCR-AA)) | 22.8 | Long, fragmented medical-record reasoning | Long-context medical reasoning | [Medical Long Context Reasoning (MLCR-AA)](https://artificialanalysis.ai/evaluations/mlcr-aa) |
| CritPt (Critical Physics Tasks) | 14.3 | Research-level physics questions | Research-level physics reasoning | [CritPt Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/critpt) |
| JevBench 1.4 | 4.77 | 534 frozen v1.2 decisions plus 308 sealed v1.4 decisions | Typed decisions with sealed generalization tests | [JevBench v1.4 method](https://github.com/fstandhartinger/jevbench/blob/5b405e7e4efff2cfbbb2fe67db9b0cbad0704b8d/docs/METHOD-v1.4.md) |
| JevBench 1.5 | 6.65 | 904 open decisions plus 720 sealed decisions per complete run | Choice, Noul, and Score decisions with sealed generalization tests | [JevBench v1.5 frozen method and disclosures](https://github.com/fstandhartinger/jevbench/blob/bb05a335bc809e61b20c0f745d25499a82b326fc/docs/METHOD-v1.5-README.md) |
| STEMBench (STEMBench120: six-domain scientific research progress) | 2.54 | 120 research tasks, 20 in each of six STEM domains | Open research targets; no run has reached a full-target success | [STEMBench: six-domain model evaluation](https://www.ulam.ai/stembench) |

### Agentic

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| Terminal-Bench 4.0 | 42.30 | 66 professional computer-work tasks | Frontier autonomous knowledge work | [Terminal-Bench 4.0](https://www.tbench.ai/news/terminal-bench-4-0) |
| AA Briefcase (Artificial Analysis Briefcase) | 1420 | Professional knowledge-work tasks | Professional work | [Artificial Analysis Briefcase Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/aa-briefcase) |
| AA AutomationBench (Artificial Analysis AutomationBench) | 68.9 | Business-process automation tasks | Agentic automation | [Artificial Analysis AutomationBench Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/automationbench-aa) |
| AA ITBench (Artificial Analysis ITBench-AA) | 46.9 | IT incident-response tasks | Enterprise IT operations | [Artificial Analysis ITBench-AA Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/itbench-aa) |
| Terminal-Bench 2.1 | 90.6 | Terminal-based software-agent tasks | Professional software engineering | [Terminal-Bench](https://www.tbench.ai/) |
| HLE w/ tools (Humanity's Last Exam with tools) | 63.9 | Expert questions with tool use | Frontier tool-augmented reasoning | [DeepSeek-V4 Technical Report](https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro/blob/main/DeepSeek_V4.pdf) |
| GDPval-AA | 1600 | Agentic real-world work tasks | Professional agentic workflows | [DeepSeek-V4 Technical Report](https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro/blob/main/DeepSeek_V4.pdf) |
| GDPval-AA (GDPval-AA normalized) | 55.0 | Economically valuable tasks | Professional agentic workflows | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/grok-4-3) |
| AA Terminal-Bench 4.0 (Artificial Analysis Terminal-Bench v4.0) | 26.8 | Terminal-based agent tasks | Agentic software engineering | [Artificial Analysis Terminal-Bench v4.0 Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/terminalbench-v4-0) |
| GDP.pdf (Artificial Analysis GDP.pdf) | 12.8 | Professional document-production tasks | Professional knowledge work | [GDP.pdf Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/gdp-pdf) |
| CyberGym | 88.1 | 1,507 vulnerability analysis instances | Real-world cybersecurity | [CyberGym: Evaluating AI Agents' Real-World Cybersecurity Capabilities at Scale](https://www.cybergym.io/) |
| ExploitGym | 15.3 | 898 exploitation tasks | Advanced cybersecurity exploitation | [ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?](https://arxiv.org/abs/2605.11086) |
| AutomationBench | 60.5 | 600 public automation tasks | Long-horizon automation | [Kimi K3: Open Frontier Intelligence](https://www.kimi.com/blog/kimi-k3) |
| Agents' Last Exam | 35.4 | Agent tasks | Advanced agentic work | [DeepSeek V4 Flash 0731 update](https://api-docs.deepseek.com/zh-cn/updates/) |
| Terminal-Bench 2.1 (Vals) (Terminal-Bench 2.1, Vals AI run) | 74.5 | Difficult terminal tasks | Frontier agentic | [Vals AI Terminal-Bench 2.1, Vals AI run leaderboard](https://www.vals.ai/benchmarks/terminal-bench-2-1) |
| CWE-bench v1 | 55.0 | Audit and patch held-out C/C++, Go, Java, TypeScript/JavaScript, Python, and Rust repositories | Frontier agent evaluation | [CWE-bench v1 leaderboard and methodology](https://cwe-bench.com/) |
| Hermes Index (Hermes Index: agentic capability and cost per task inside Hermes Agent) | 36.9 | Four suites: Hermes Bench (150 tasks), TerminalBench 4 without its 4 GPU tasks, TerminalBench Science and SkillsBench | Frontier agentic | [Hermes Index: an agentic model leaderboard](https://portal.nousresearch.com/bench) |
| Hermes Bench (Hermes Bench: 150 agent tasks graded on the files and state the agent leaves behind) | 70.7 | 150 tasks across 25 categories | Frontier agentic | [Hermes Bench (Hermes Index)](https://portal.nousresearch.com/bench) |
| MazeBench (MazeBench: visual spatial reasoning in a 3D open world) | 1 | 100 hidden gems across 256 rooms in one open world | Long-horizon spatial planning, with single puzzles that can take over one hundred moves | [MazeBench Results](https://mazebench.com/blog?post=maze-bench-results) |
| CWE-bench (CWE-bench v2) | 57.6 | Audit and patch C/C++, Go, Java, TypeScript/JavaScript, Python, and Rust repositories; 100 public and 150 held-out tasks | Frontier agent evaluation | [CWE-bench v2 leaderboard and methodology](https://cwe-bench.com/) |
| Terminal-Bench 4.1 | 38.50 | 66 professional computer-work tasks | Frontier autonomous knowledge work | [Terminal-Bench 4.1](https://www.tbench.ai/news/terminal-bench-4-1) |
| AA Cyber Index (Artificial Analysis Cyber Index) | 40.8 | Three defensive cybersecurity evaluations | Real-world vulnerability discovery and patching | [Artificial Analysis Cyber Index](https://artificialanalysis.ai/evaluations/artificial-analysis-cyber-index) |
| CWE-Bench-AA (Artificial Analysis CWE-Bench-AA) | 68.3 | Real-world vulnerability patching tasks | Automated vulnerability remediation | [CWE-Bench-AA Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/cwe-bench-aa) |
| DeepsecBench-AA (Artificial Analysis DeepsecBench-AA) | 31.2 | Vulnerability discovery in real codebases | Security code review | [DeepsecBench-AA Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/deepsecbench-aa) |
| CyberGym-E2E-AA (Artificial Analysis CyberGym-E2E-AA) | 22.9 | 131 end-to-end vulnerability reproduction and patching tasks | Real-world cybersecurity | [CyberGym-E2E-AA Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/cybergym-e2e-aa) |

### Multimodal & Grounded

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| Chartography (tools) (Chartography with image and code tools) | 83.0 | 100 specialized chart types | Professional chart reasoning | [Chartography](https://surgehq.ai/blog/chartography) |
| AA-MMMU-Pro (Artificial Analysis MMMU-Pro) | 77.0 | Multimodal academic reasoning | Frontier multimodal | [Artificial Analysis MMMU-Pro Benchmark Leaderboard](https://artificialanalysis.ai/evaluations/mmmu-pro) |
| Design Arena Website (Design Arena Website Elo) | 1289 | Website generation comparisons | Design and website generation | [OpenRouter Grok 4.3 benchmarks](https://openrouter.ai/x-ai/grok-4.3/benchmarks) |
| BabyVision w/ Python (BabyVision with Python) | 92.5 | Visual perception tasks with Python | Fine-grained visual perception | [Kimi K3: Open Frontier Intelligence](https://www.kimi.com/blog/kimi-k3) |
| ZeroBench w/ Python (ZeroBench_main with Python) | 51.0 | Visual reasoning questions with Python | Tool-augmented visual reasoning | [Kimi K3: Open Frontier Intelligence](https://www.kimi.com/blog/kimi-k3) |
| ParseBench-1 (ParseBench-1: document parsing for AI agents (LlamaIndex leaderboard, first release)) | 64.38 | 2,078 human-verified enterprise pages from insurance, finance and government documents | No method is strong on all five dimensions; charts and visual grounding separate the field | [ParseBench: A Document Parsing Benchmark for AI Agents](https://arxiv.org/abs/2604.08538) |
| ExtractBench (LlamaIndex) (ExtractBench: schema-guided enterprise document extraction (LlamaIndex leaderboard)) | 87.11 | 370 documents, 4,869 pages, 8 business domains, 67 document types, each with a JSON Schema | Short documents are near solved; long record lists and grounding separate systems | [ExtractBench: A Benchmark for Schema-Guided Enterprise Document Extraction](https://arxiv.org/abs/2607.29677) |

### external

| Benchmark | Value | Tasks | Difficulty | Provenance |
| --- | --- | --- | --- | --- |
| SEC-Bench Pro | 62.8 | Security engineering tasks | Advanced cybersecurity | [Introducing GPT-5.6](https://openai.com/index/gpt-5-6/) |
| Cookie-Bench (Cookie-Bench: Continuous On-screen Key Interaction Evaluation for Web Generation) | 87.9 | 1,000 web development queries (React setting) | Interactive web apps and static pages across three tiers | [Cookie-Bench: Continuous On-screen Key Interaction Evaluation for Web Generation](https://arxiv.org/abs/2605.30000) |
| BlenderWorld (text-only) (BlenderWorld: Benchmarking Visual Agents for 3D Creation from Real-World References (text-only subset)) | 79.2 | 78 text-brief 3D creation tasks in Blender | Long-horizon 3D modeling and scene building | [BlenderWorld: Benchmarking Visual Agents for 3D Creation from Real-World References](https://arxiv.org/abs/2610.23986) |
| MaxxBench (MaxxBench (mutated DeepSWE tasks)) | 55.6 | 89 mutated DeepSWE feature-implementation tasks, two attempts per model | Long-horizon software engineering under changed requirements | [Why MaxxBench (unLAP research post)](https://maxxbench.unlap.ai/research/why-maxxbench) |
| Chess Puzzles (Chess Puzzles (Epoch AI)) | 46.0 | 100 programmatically generated chess positions with a single best move | Novel chess positions, single best move | [Chess Puzzles (Epoch AI)](https://epoch.ai/benchmarks/chess-puzzles) |
| Mystery Game Puzzles (Mystery Game Puzzles (Epoch AI)) | 49.0 | 100 single-best-move positions from an undisclosed game | Unfamiliar game positions, single best move | [Mystery Game Puzzles (Epoch AI)](https://epoch.ai/benchmarks/mystery-game-puzzles) |
| Furniture Assembly (Furniture Assembly Benchmark (Epoch AI)) | 34.2 | 60 photos of three IKEA builds, some with deliberate assembly mistakes | Spatial reasoning from technical diagrams to real photos | [Furniture Assembly Benchmark (Epoch AI)](https://epoch.ai/benchmarks/furniture-assembly) |
| MirrorCode (MirrorCode (Epoch AI and METR)) | 64.4 | 30 reimplementation tasks (15 programs in two languages), three attempts each | Days-long software engineering without source code | [MirrorCode (Epoch AI)](https://epoch.ai/benchmarks/mirrorcode) |
| Landing Page Arena | 1504 | 12 client briefs for small-business landing pages, rated by human votes | Marketing page design and copy | [BenchLM](https://benchlm.ai/benchmarks/landing-page-arena) |
| Landmark Arena | 1411 | 20 famous landmarks built as three.js web pages, rated by human votes | 3D scenes in browser code | [BenchLM](https://benchlm.ai/benchmarks/landmark-arena) |
| NYT Connections (NYT Connections LLM Benchmark (extended version)) | 92.8 | 940 NYT Connections puzzles, each with up to four added trick words | Wordplay, word knowledge and decoys that fit almost-but-not-quite | [NYT Connections LLM Benchmark (Lech Mazur, GitHub)](https://github.com/lechmazur/nyt-connections) |
| StealthMark (StealthMark Intelligence Benchmarks (overall score)) | 59.1 | 8 original tasks: Newton's Mirror, Ground Truth, Latch, Foresight, Mona Lisa, Selfie, Beach and Raccoon | Chained physical reasoning, planning without tools and precise visual construction | [StealthMark: The AI Benchmark Explained (Stealth Models)](https://stealthmodels.com/stealthmark) |
| FrontierPhysics (End-to-End Physics Research Benchmark for AI Agents) | 3.6 | 56 research projects from 42 physicists across nine fields, three trials each (168 rollouts per setup) | Open-ended research projects that took their authors about 270 days on average | [FrontierPhysics: Evaluating Agents for End-to-End Physics Research (arXiv, November 4, 2026)](https://arxiv.org/abs/2611.06910) |
| HoneyBench (HoneyBench v0.2 (Goodhart Labs)) | 76.4 | 20 private honeypot tasks across 7 domains with exploitable evaluators | Agentic alignment and specification gaming | [HoneyBench v0.2](https://goodhartlabs.com/honeybench) |

## Lifecycle and limitations log

Lifecycle events the source associates with this model.

No lifecycle event references this model. That is not evidence the model has no lifecycle plan — only that this source published none.

## What this profile does not claim

Values are reproduced exactly as their sources published them, in the units those sources declared; none are converted, interpolated, or averaged across providers. Any field marked unavailable was attempted and not returned. Last attempted fetch for this model's score: 2026-11-21 05:17 UTC.
