Economic work
We track 91 benchmarks in economic work, or 101 if you count each version. They run from everyday level to expert level.
Anthropic held the top economic work score 59% of the contested time
That's about 27 of the 46 months since January 2023, averaged over the benchmarks below. The easiest benchmarks are at the top. Versions of one test share a row, with a dark tick at each new version. We stop the clock, and fade the rest of the row, when the best score passes 85%, or at the newest score if 45 days go by without another.
Anthropic Google DeepMind DeepSeek OpenAI ByteDance Seed Z.ai Tencent Meta
University Undergraduates and entry-level professionals
Expert Specialists with years of training, often over hours of work
Share of contested time at number one
Anthropic 59% OpenAI 26% 
DeepSeek 5% 
ByteDance Seed 4% 
Meta 3% 
Google DeepMind 3% Z.ai 1%
DeepSeek 5% ByteDance Seed 4% Meta 3% Google DeepMind 3% Z.ai 1% Closed out: 2 benchmarks passed 85%
ByteDance Seed 1 Anthropic 1 - Jun 2026 GDPval Seed2.1 Pro
- Apr 2026 OfficeQA Claude Opus 4.7
Scores come from each benchmark’s own leaderboard, from Epoch AI’s runs and from labs’ reports on their own models. We date Epoch runs by the model’s release and lab reports by when they came out. Labs pick their own settings, so a lab’s claim can beat an independent run of the same model. Turn off lab claims to see independent runs only.
Anthropic Google DeepMind DeepSeek OpenAI Moonshot AI Tencent Z.ai Meta
University Undergraduates and entry-level professionals
Expert Specialists with years of training, often over hours of work
Share of contested time at number one, independent runs only
Anthropic 65% OpenAI 19% DeepSeek 6% Moonshot AI 5% Google DeepMind 4% Meta 1%
DeepSeek 6% Moonshot AI 5% Google DeepMind 4% Meta 1% Only each benchmark’s own leaderboard and Epoch AI’s runs, dated by the model’s release. We leave out benchmarks that only labs report.
From everyday to expert level
Grouped by who could solve the tasks, from things most adults can do to open research problems. The bar under each name is its best score.
Everyday
Most adults, without training
- Vending-Bench 2
- FoodTruck Bench
- Vending-Bench
- Information Extraction (ByteDance Seed in-house)
- Cohere enterprise human evaluation
- Information Processing (ByteDance Seed in-house)
- Intention Recognition (ByteDance Seed in-house)
- Compositional Tasks (ByteDance Seed in-house)
- Enterprise generative evaluation (Cohere)
- Online Experience (Moonshot)
- Text Classification (ByteDance Seed in-house)
- DesignArena HTML Slides
- DesignArena Proprietary Harness Slides
- DesignArena Python-PPTX Slides
- DesignArena Spreadsheet
- Genspark Slides Benchmark
University
Undergraduates and entry-level professionals
Expert
Specialists with years of training, often over hours of work
- ProfBench
- APEX-Agents
- Agents' Last Exam
- GDPval-AA v2
- JobBench
- Vals Finance Index
- Vals Legal Index
- Vending-Bench Arena
- GDPval
- GDPval-AA
- Clipping Bench
- Remote Labor Index
- GDPval-AA v2.1
- $OneMillion-Bench
- CommerceAgentBench
- Holdings-IndividualTaxBench
- AA-Briefcase v1.1
- CoWorkBench
- SpreadsheetBench 2
- AA-Briefcase
- OR-License
- EnterpriseClawBench
- MissionReport
- Short Form Bench
- xbench-Recruitment
- xbench-Marketing
- Epoch Automation Reports
- XpertBench
- E-Bench (Tencent)
- Ko-GDPval
- Agent Startup Bench
- Agents' Last Exam v1.1
- Business Operation (Moonshot)
- E-Bench-Code (Tencent)
- GDPval-MM
- GDPval-Rubrics (MiniMax)
- Kimi Claw 24/7 Bench
- 24/7 ClawBench 2.0
- DECK Bench
- DeepSeek Chinese white-collar tasks evaluation
- GDPVal-Diamond
- MIRA Bench
- AA-AnalystAgent
- AlphaEval
- APEX-1
- APEX-Accounting
- APEX-Agents 1.1
- BizBench
- Data Science Tasks (OpenAI)
- DAYJOB: Finance
- Design Tasks (OpenAI)
- Digital Work Index
- EnterpriseBench: CoreCraft
- GDP.xlsx
- GDPevo
- GDPval Extended
- Management Consulting Tasks (OpenAI)
- MerchantBench
- MMEnterprise-Bench
- OccuBench
- SnorkelRevOps
- SOP-Bench
- Tuesday Work Index
- WildArtifactBench
- WorkplaceAgents
Another 4 have no level. Levels don’t apply to arenas, safety evaluations or composite indices.
By kind
Tasks from real jobs
| Benchmark | Level | Scored | Best score | Status |
|---|---|---|---|---|
| ProfBench Model writes expert-level reports in physics, chemistry, finance and consulting. Calibrated LLM judges grade against PhD/MBA-written rubrics. | Expert | Graded by a model | 66.2% | In use |
| Agents' Last Exam 1,500+ expert-sourced tasks from real projects across 55 occupations in three difficulty tiers. Objective verifiers grade deliverables. 5-hour cap. | Expert | Checked automatically | 45.9% | In use |
| GDPval-AA v2 Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Outputs are compared pairwise and ranked by Elo. | Expert | Graded by a model | 1,861 | In use |
| JobBench 130 tasks from 35 occupations workers most want delegated. Agent works through messy reference files, graded by fact-anchored binary rubrics. | Expert | Graded by a model | 61.4% | In use |
| GDPval Real work tasks from 44 occupations with reference files. Experts blindly compare model deliverables with expert work. Metric is win-or-tie rate. | Expert | Graded by people | 87.9% | In use |
| GDPval-AA Agents complete GDPval occupational tasks. Artificial Analysis compares the outputs in pairs and ranks models by Elo. | Expert | Graded by a model | 1,932 | Fading |
| Remote Labor Index Agents complete real paid remote freelance projects from many sectors. Reviewers judge whether each deliverable would be accepted as commissioned work, giving an automation rate. | Expert | Graded by people | 21.3% | Not in cards |
| GDPval-AA v2.1 Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Blind pairwise comparisons give Elo ratings. | Expert | Graded by a model | 1,846 | In use |
| $OneMillion-Bench Agent solves 400 expert-curated tasks in law, finance, industry, healthcare and science. Rubric grading of accuracy, logic, feasibility and compliance. | Expert | Graded by a model | 42.1% | In use |
| Thrive Holdings helpdesk ticket resolution benchmark Agent investigates IT support requests, takes action and decides when to escalate to a technician. Scored by pass rate. | University | Checked automatically | 61.5% | Not in cards |
| Holdings-IndividualTaxBench 88 real client document packages turned into Form 1040 fields. Fact-level accuracy against practitioner-reviewed returns, with an LLM judge matching values. | Expert | Checked automatically | 76.3% | Not in cards |
| AA-Briefcase v1.1 Version 1.1 of Artificial Analysis's multi-hour office work evaluation, with Elo ratings fitted by Crowd-BT. | Expert | Graded by a model | 1,822 | In use |
| CoWorkBench Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished). | Expert | Graded by a model |  75.3% | In use |
| AA-Briefcase Artificial Analysis knowledge-work evaluation of professional deliverables. Models ranked by Elo. | Expert | Graded by a model | 1,720 | In use |
| OR-License Agents turn original client materials into validated operations research solutions without an expert first reformulating the problem. | Expert | Graded by people | 88.4% | Not in cards |
| XpertBench Models complete expert tasks from more than 140 task types in law, finance, education, humanities and science. Over 200 domain experts wrote the prompts and weighted rubrics. | Expert | Graded by a model | 64.5% | Fading |
| Ko-GDPval Agents complete 170 Korean office tasks from 58 occupations in 11 industries and produce deliverable files. Rubrics grade data, content, structure, quality and format. | Expert | Graded by a model |  87.3% | In use |
| Agents' Last Exam v1.1 Agents complete 122 public long-horizon professional workflow tasks across 55 sub-industries. Deterministic graders check outcomes against hidden references. | Expert | Checked automatically | 77.4% | In use |
| GDPval-MM Agents produce office deliverables for GDPval tasks in a MiniMax agent framework. A judge model compares full trajectories pairwise and reports a win rate. | Expert | Graded by a model |  59% | Fading |
| GDPval-Rubrics (MiniMax) MiniMax rubric grading of model deliverables on public GDPval tasks, in an environment matched to GDPval-AA. | Expert | Graded by a model | 74.8% | In use |
| Kimi Claw 24/7 Bench Agents work on 17 persistent multi-day professional scenarios through OpenClaw, with 610 checkpoints. The score is the mean pass rate over three runs. | Expert | Checked automatically | 46.9% | In use |
| 24/7 ClawBench 2.0 Agents do always-on assistant work through OpenClaw. Tasks span several days, events arrive at the same time, and interruptions are routine. | Expert | Checked automatically | 48.3% | System cards only |
| GDPVal-Diamond Models complete a ByteDance Seed subset of GDPval occupational tasks. Rubrics grade each deliverable automatically. | Expert | Graded by a model | 21.3% | System cards only |
| APEX-1 Models complete expert-written work cases with source documents for four jobs: investment banking, consulting, big law and primary care. A judge model grades each response against rubrics. | Expert | Graded by a model | Not in cards | |
| DAYJOB: Finance Agents get short requests from finance colleagues plus a workspace of filings, ledgers and emails. An agent judge applies an expert rubric, and a pass needs 95% of criteria. | Expert | Graded by a model | Not in cards | |
| GDPval Extended Models produce work products for occupations across nine GDP sectors. Mercor adds 132 new expert-written tasks to the 213 public ones and reports both scores. | Expert | Graded by a model | Not in cards | |
| OccuBench Agents do 100 professional task scenarios across 10 industries and 65 domains. Language world models simulate each domain's tools so tasks stay solvable. | Expert | Checked automatically | Not in cards | |
| WildArtifactBench Agents produce real deliverables in any format, from audio analysis and medical video reads to 3D meshes and kitchen layouts. Human and agent judges compare them with a baseline agent's work to give win rates and Elo. | Expert | Ranked by votes | Not in cards | |
| WorkplaceAgents Agents produce finished work products for 200 professional tasks across 96 occupations in 19 sectors. Harbor grades deliverables with code checks and rubrics. | Expert | Graded by a model | Not in cards |
Running a business
| Benchmark | Level | Scored | Best score | Status |
|---|---|---|---|---|
| Vending-Bench 2 Agent runs a simulated vending-machine business for one year starting with $500. Score is the final bank balance, averaged over runs. | Everyday | Checked automatically | $16,484 | Fading |
| E-Commerce Bench Agents start with 100,000 yuan and run up to four online stores for 365 simulated days on real market data. They source, negotiate with suppliers, price, promote and manage cash. Scored by end-of-year assets over the opening balance. | University | Checked automatically | 14.3 | Not in cards |
| CommerceAgentBench Agents finish real e-commerce work across procurement, listings, operations, fulfillment and after-sales through CLI, browser, file and API tools. A task passes only if every verifier check on the end state succeeds. | Expert | Checked automatically | 61.7% | Not in cards |
| Vending-Bench Agents run a simulated vending machine business over long horizons, ordering stock, setting prices, and paying daily fees. Scoring is final net worth in dollars. | Everyday | Checked automatically | $4,694 | Dropped |
| xbench-Recruitment Agents do 50 headhunting tasks from live business cases: company mapping, finding facts about people and finding people who fit a profile. A judge scores answers 1 to 5. | Expert | Graded by a model | 78.5 | Not in cards |
| xbench-Marketing Agents search for influencers who fit 50 real advertiser briefs, scored against a curated pool of 836 influencers. | Expert | Graded by a model | 50.8 | Not in cards |
| Agent Startup Bench Agents complete workflows taken from market-validated AI-native startup products and their users. Domain experts wrote the tasks, attachments and rubrics after user interviews. | Expert | Graded by a model | 68.8% | In use |
| Business Operation (Moonshot) Agents manage a simulated business over a long horizon and make operating decisions. Moonshot reports a percentage score. | Expert | Settled by events | 37.8% | In use |
| Cohere enterprise human evaluation Trained annotators blindly compare two models' answers on enterprise tasks such as HR FAQs, technical FAQs and workplace assistants. They judge accuracy, instruction following and style. Results are win, tie and loss rates. | Everyday | Ranked by votes |  51.7% | Dropped |
| Intention Recognition (ByteDance Seed in-house) Models classify intent from dialogues, meeting transcripts, social posts and articles, such as purchase intent in customer service chats. | Everyday | Checked automatically | 63.2% | Fading |
| YC-Bench Agent runs a simulated startup for one year via CLI, managing staff, clients and cash. Scored by final funds. | University | Checked automatically | $2,100,000 | In use |
| Compositional Tasks (ByteDance Seed in-house) Models identify several elements of a business text in one inference, such as user intent and slot values. | Everyday | Checked automatically | 59.1% | System cards only |
| Enterprise generative evaluation (Cohere) Models complete 22 enterprise writing tasks such as job ads with format rules. Rule checks and LLM judges give a pass rate. | Everyday | Graded by a model | 94.2% | System cards only |
| Text Classification (ByteDance Seed in-house) Models analyze business text and return labels or structured outputs, for example sentiment or the main viewpoints in user content. | Everyday | Checked automatically | 69% | System cards only |
| AA-AnalystAgent Agents answer 80 numeric analyst questions from real spreadsheets and documents in 14 domains, using Python in a sandbox. A question counts only if all five attempts are right. | Expert | Graded by a model | Not in cards | |
| AlphaEval Agent products such as Claude Code and Codex do 94 production tasks from 7 companies across 6 O*NET domains. Grading mixes formal checks, UI tests and rubrics. | Expert | Graded by a model | Not in cards | |
| BizBench Long-horizon agents run business operations such as payments under long, partly outdated or conflicting policies. Intermediate steps are checked, not only final success. | Expert | Checked automatically | Not in cards | |
| GDPevo Agents learn from training tasks drawn from enterprise workflows and must reuse that experience on related test tasks. Gains at test time measure self-evolution. | Expert | Checked automatically | Not in cards | |
| Management Consulting Tasks (OpenAI) Models do management consulting work tasks from OpenAI's internal set. OpenAI has not published the task format or grading. | Expert | Graded by a model | Not in cards | |
| SnorkelRevOps Agents work 200 tasks in a simulated B2B SaaS revenue team spanning CRM, CPQ, billing, compensation, marketing and finance records. | Expert | Graded by a model | Not in cards | |
| SOP-Bench Agents follow expert-written standard operating procedures across 12 industrial domains, using executable tool interfaces over more than 2,000 tasks with 10 to 50 decision points. | Expert | Checked automatically | Not in cards |
Documents and spreadsheets
| Benchmark | Level | Scored | Best score | Status |
|---|---|---|---|---|
| FORTE Agents complete 180 office tasks across 15 corporate professions inside a workspace of xlsx, docx, pptx, pdf and other files. An LLM judge checks the output files against expert rubrics, and a task passes only if every rubric item passes. | University | Graded by a model | 77.8% | In use |
| OfficeQA Pro Models answer the 133 hardest OfficeQA questions by finding tables in U.S. Treasury Bulletin documents and computing exact numeric answers. | University | Checked automatically | 80.6% | In use |
| OfficeQA Grounded reasoning over a large corpus of U.S. Treasury Bulletin documents: locate tables and compute precise numeric answers. | University | Checked automatically | 86.3% | In use |
| SpreadsheetBench Real spreadsheet manipulation questions from Excel forums. Output workbook is checked cell-by-cell against reference solutions. | University | Checked automatically | 89.3% | In use |
| SpreadsheetBench 2 Agents complete 321 expert-annotated business spreadsheet workflows, such as financial modeling, debugging, and charting, in multi-sheet workbooks. Scoring is task accuracy. | Expert | Checked automatically | 34.9% | In use |
| EnterpriseClawBench Agents do 852 tasks rebuilt from real proprietary workplace agent sessions, each with fixtures, hard rules and semantic rubrics. | Expert | Graded by a model | 66.3% | Not in cards |
| MissionReport Models write government status reports in Word from working files and a template, across 20 expert-written Department of Defense style assignments. Expert rubrics score each decision. | Expert | Graded by a model | 64.1% | Not in cards |
| WorkSpaceBench Agent completes 388 tasks in realistic worker file workspaces (20k files) needing cross-file dependency reasoning. Scored against 7,399 rubrics. | University | Graded by a model | 67.7% | In use |
| PresentBench Models generate full slide decks from background materials for 238 real-world cases. A judge scores each deck against about 54 hand-written checklist items. | University | Graded by a model | 79.6% | In use |
| SpreadsheetBench Verified Models solve 400 real spreadsheet manipulation questions from Excel forums, an expert annotated subset of SpreadsheetBench. Each answer runs against several test spreadsheets. | University | Checked automatically | 79.1% | Fading |
| E-Bench (Tencent) Agents complete Tencent in-house work tasks in the working agent group. Task and scoring details are not published. | Expert | Checked automatically |  77.1% | In use |
| Information Extraction (ByteDance Seed in-house) Models extract words, sentences or fields from meeting records, legal texts, contracts, product descriptions and knowledge bases. | Everyday | Checked automatically | 52% | Fading |
| DECK-Arena Models build presentations for real user queries with python-pptx, pptxjs or a Kimi slide language, with search and image generation. An automated grader rates them as Elo scores. | University | Graded by a model | 1,456 | In use |
| E-Bench-Code (Tencent) Agents complete Tencent in-house work tasks that need code. Task and scoring details are not published. | Expert | Checked automatically | 79% | In use |
| Information Processing (ByteDance Seed in-house) Models analyze large collections of user-generated content, for example summarizing viewpoints and judging overall sentiment. | Everyday | Graded by a model | 53.9% | Fading |
| MEWC Models solve 179 Microsoft Excel World Championship problems from 2021 to 2026 with Excel tools. Output cells are compared one by one with answer cells. | Competition | Checked automatically | 74.4% | Fading |
| MEWC v2 Models solve 100 hard Microsoft Excel World Championship problems with Excel tools. Expert rubrics score the answers over three trials. | Competition | Graded by a model | 63.3% | System cards only |
| OfficeQA Pro V2 Agents answer 90 analytical questions using about 120,000 pages of U.S. Treasury Accounts of Receipts and Expenditures from 1793 to 2024. Answers are scored by exact match. | University | Checked automatically | 64.4% | In use |
| DECK Bench Models produce presentation decks from task descriptions drawn from real usage. The score measures deck quality. | Expert | Graded by a model | 73.5% | System cards only |
| DeepSeek Chinese white-collar tasks evaluation Agents complete 30 Chinese professional analysis, document generation and editing tasks across 13 industries. Blind human annotators compare outputs against a baseline model. | Expert | Ranked by votes | 53% | System cards only |
| Online Experience (Moonshot) Agents produce the deliverable file types that real online users request most often, following the distribution of real usage. | Everyday | Graded by a model | 77.9% | System cards only |
| Cohere North internal evals Cohere internal evals of its North agent platform: agentic question answering, spreadsheet data analysis and memory use. LLM-as-judge scoring. | University | Graded by a model | Not in cards | |
| Data Science Tasks (OpenAI) Agents do data science and analysis tasks from an OpenAI internal set. OpenAI has not published the task count or grading. | Expert | Graded by a model | Not in cards | |
| Design Tasks (OpenAI) Models produce design work such as slides and documents that follow templates, from an OpenAI internal set. OpenAI has not published the grading. | Expert | Graded by a model | Not in cards | |
| DesignArena HTML Slides Agents write a self-contained HTML slide deck that is converted to PPTX. Users compare the decks and vote for the better one. | Everyday | Ranked by votes | Not in cards | |
| DesignArena Proprietary Harness Slides Slide-making products build a presentation from a user prompt in their own harness. Users compare the decks and vote for the better one. | Everyday | Ranked by votes | Not in cards | |
| DesignArena Python-PPTX Slides Agents build PowerPoint decks with python-pptx from a user prompt. Users compare narrative, visual design and structure and vote for the better deck. | Everyday | Ranked by votes | Not in cards | |
| DesignArena Spreadsheet Agents build spreadsheets such as financial models, budgets and trackers as Excel or CSV workbooks. Users compare accuracy, structure and formatting and vote. | Everyday | Ranked by votes | Not in cards | |
| GDP.xlsx Agents open a professional Excel workbook in a sandbox and answer a short question about it, as a domain professional would. A Gemini autograder checks the final answer against criteria. | Expert | Graded by a model | Not in cards | |
| Genspark Slides Benchmark Models build finished slide decks from 200 real production tasks in the Genspark Slides agent harness. An internal grader scores quality, completion, content and design. | Everyday | Graded by a model | Not in cards |
Simulated companies
| Benchmark | Level | Scored | Best score | Status |
|---|---|---|---|---|
| APEX-Agents 480 long-horizon tasks in banking, consulting and law worlds with docs, sheets, email and chat. Rubric-graded, Pass@1. | Expert | Graded by a model | 82.4% | In use |
| Vending-Bench Arena Several agents each run a vending machine at the same location in the Vending-Bench 2 environment. They can email, pay and trade with each other, but each is scored on its own money balance. | Expert | Ranked by matches | $15,962 | Not in cards |
| Clipping Bench Agents run a simulated clip farm for about a month. They join pay-per-view campaigns, cut clips from licensed streams, post them and compete with rival clippers for budgets. | Expert | Settled by events | $9,514 | Not in cards |
| TheAgentCompany Agents do work tasks in a simulated software company, using a browser, code, and chat with simulated coworkers. Scored by full and partial checkpoint completion. | University | Checked automatically | 42.9% | Dropped |
| FoodTruck Bench Agents run a food truck in Austin, Texas for a 30-day simulation, choosing locations, menus, prices, inventory, staff and loans with 34 tools. Humans can play the same game. | Everyday | Settled by events | $75,264 | Not in cards |
| Short Form Bench Agents act as growth lead of a small brand for a simulated week. They post short videos, run the store and answer customers, sponsors and the founder. | Expert | Settled by events | $2,949 | Not in cards |
| Epoch Automation Reports Agents in coding harnesses do real Epoch AI work tasks such as graphic design, data insights and research design. Epoch staff grade outputs against the rubric they use for employees. | Expert | Graded by people | 65.3% | Not in cards |
| MIRA Bench Agents carry out long multi-role enterprise tasks across several systems and decide when to delegate to subagents. | Expert | Checked automatically | 64.1% | System cards only |
| APEX-Accounting Agents run month-end close tasks across 10 frozen business worlds with the apps an accountant uses. 100 tasks are graded against 1,475 rubric criteria. | Expert | Graded by a model | Not in cards | |
| APEX-Agents 1.1 Agents do long banking, consulting and law tasks in worlds with docs, sheets, email and chat. A new judge scores hedged multiple answers as zero. | Expert | Graded by a model | Not in cards | |
| EnterpriseBench: CoreCraft Agents work inside a simulated hardware startup with more than 2,500 entities and 23 tools. They triage incidents, audit shipments and handle customers under company policy. | Expert | Checked automatically | Not in cards | |
| MerchantBench Agents run a seller's e-commerce store for 365 simulated days at order level. Actions constrain later choices, so the test measures long-term coherence. | Expert | Settled by events | Not in cards | |
| MMEnterprise-Bench Enterprise agents answer 1,613 questions over workplace documents, chats and figures that change over time. They must resolve conflicting versions and say when evidence is not enough. | Expert | Checked automatically | Not in cards |
Composite work indices
| Benchmark | Level | Scored | Best score | Status |
|---|---|---|---|---|
| Vals Finance Index Agents are scored on Finance Agent v2.1, Excel Modeling Benchmark v1.1 and Tax Agent Bench, combined with weights from each sub-sector share of US GDP. | Expert | Graded by a model | 67% | Not in cards |
| Vals Index v2.2 Agents are scored on eight finance, coding, legal, and tax benchmarks, combined with weights from each sector's share of US GDP. | Graded by a model | 70.8% | In use | |
| Vals Legal Index Agents are scored on Legal Research Bench, HLAB-Vals-1.1 and Tax Agent Bench, mapped to practice areas weighted by share of US law-office revenue. | Expert | Graded by a model | 77.9% | Not in cards |
| Vals Index v2.1 Version 2.1 of Vals AI's GDP-weighted index, which moved coding to Terminal-Bench 4.0 and added a tax sector. | Graded by a model | 68.9% | In use | |
| Vals Index v1 Vals AI's first index of agentic work in finance, law and coding, with each sector weighted by its share of U.S. GDP. | Graded by a model | 74.7% | System cards only | |
| Digital Work Index Models are scored on eight benchmarks in coding, reasoning, writing and enterprise work. Weights come from the wages of the 29 digital occupations the benchmarks map to. | Expert | Graded by a model | Not in cards | |
| Tuesday Work Index A composite score built from the Surge AI benchmark suite, such as DAYJOB, GDP.pdf, Riemann-bench and ComplexConstraints. It asks whether models get through an ordinary workday. | Expert | Graded by a model | Not in cards | |
| Vals Index v2 Version 2 of Vals AI's GDP-weighted index, which brought back a legal sector and added financial modeling in Excel and code migration. | Graded by a model | Not in cards |