GPT-5.6 Sol: A Revolution in Financial Management with ML Model

A Technological Advancement in Finance with GPT-5.6 Sol

Model ML has integrated GPT-5.6 Sol into its processes to enhance the efficiency of financial workflows. This model allows for the creation of editable PowerPoint and Excel files while reducing the number of tokens per presentation by 21% compared to Fable 5. This technological advancement marks a significant step in optimizing repetitive and time-consuming tasks in the financial sector.

Promising Results

The performance of GPT-5.6 Sol translates into several efficiency gains: a 21% reduction in tokens for PowerPoint presentations compared to Fable 5, and a 16.6 percentage points lead in professional readiness compared to Opus 5. Additionally, the model uses 36% fewer tokens for Excel spreadsheets compared to Opus 5, and enables the creation of a custom data sheet in just 5 minutes, down from about an hour previously.

In preparing a financial analysis for clients or decision-makers, teams must complete a crucial final step: reconciling evidence, formatting files, verifying each figure, and ensuring that every claim is well-sourced. The final product, whether a PowerPoint or an Excel spreadsheet, must be editable and ready for review.

Model ML co-founders Arnie and Chaz Englander recognized the extent of the work required when they began investing through a private family office after two successful exits. To address this need, they developed software that has evolved into an essential tool for finance professionals.

Automation of Financial Workflows

Model ML offers complete automation of financial workflows, from the initial request to research, analysis, and the production of final documents. A lead agent coordinates the work, selects the appropriate tools, reconciles evidence, and performs calculations, directing each step to the most suitable model, often GPT-5.6 Sol. Model ML's documentation tool allows for the creation of native PowerPoint and Excel files with traceable sources.

Model ML describes its product as “surface-agnostic,” enabling finance professionals to start a task via email or through the Model ML app and continue in their Microsoft Office plug-ins without having to re-explain everything.

Improving the Quality of Deliverables

Model ML has evaluated GPT-5.6 Sol by comparing it to other leading models across various financial workflows. Composite assessments track a mission from the initial financial brief to the creation of an editable deck or spreadsheet, verifying figures, sources, formulas, and structure, as well as the visual quality of presentations.

For PowerPoint, Model ML's Composite assessment incorporates real workflows and covers hundreds of generated presentations, with a detailed scoring grid. GPT-5.6 Sol completed the PowerPoint workflow in 100% of test cases, compared to 76% for Opus 5, and surpassed Model ML's professional readiness threshold in 43.3% of cases, compared to 26.7% for Opus 5.

Benchmarks for PowerPoint and Excel Creation

Model ML's benchmarks show that GPT-5.6 Sol achieved an overall score of 59.9% for deck quality, with a professional readiness rate of 43.3%. The model generated PowerPoint decks in 100% of cases, with adherence to the brief at 78.8% and visual quality at 77.9%. In terms of efficiency, the model used 1.10 M tokens per deck.

For Excel, the accuracy of key results reached 83.3%, with 50% of models being fully correct. The model found results in 100% of cases, with a workbook contract also at 100%. The average time per workbook is 7 minutes, with a usage of 2.44 M tokens per workbook.

Towards a New Era of Financial Work

The Composite results have convinced Model ML to expand the use of GPT-5.6 Sol in production, including for some workflows previously managed by Opus 4.8. For PowerPoint workflows, GPT-5.6 Sol combines competitive presentation quality with a higher rate of review-ready presentations than Opus 5 and Fable 5, while using about 21% fewer tokens than Fable 5.

Investment Presentations Ready for Review

The PowerPoint test allowed Model ML to verify whether GPT-5.6 Sol was ready for real work. A deck may appear polished but fail in review if its figures are incorrect or untraceable, if its graphics are flattened, or if its slides need to be rebuilt before they can be shared.

Model ML executed the model within its agent, paired with document creation and editing tools that allow the agent to create slides with editable graphics and tables. It keeps the original brief in context while working, then visually reviews each slide before returning the file.

The Evolution of Financial Work

Model ML's clients are moving towards browser-based results that remain connected to the models and source material. The Model ML platform can generate secure and interactive results that can be continuously updated or locked at a specific point in time. A reviewer could open an investment summary, click to access the financial model behind a figure, and work with the agent from the same page.

Arnie Englander, co-founder of Model ML, emphasizes that “PowerPoint, Excel, and Word were designed for a world where the creation of intellectual work was manual. AI has changed that assumption. The software itself is about to change.”