# Neuron release history

The supported Excel functions can be found in [Neuron function support](/build-spark-services/neuron/neuron-function-support.md).

## `1.27.0` 2026-03-30

* Improvements to Data Table handling when column headers are repeated.
* Fixes to Lambda helper functions `MAKEARRAY()`, `MAP()`, `REDUCE()`, `SCAN()` to use relative cell references correctly.
* Function enhancements and fixes: `SORTBY()`.
* Performance enhancements and bug fixes.

## `1.26.0` 2026-02-03

* Implementation of the `LONGTEXT()` and  `SCAN()` functions.
* Fixes for `LAMBDA()` functions.
* Fixes for Excel Data Table calculations.
* Fixes for `Xrichoutput` being inconsistent with prior versions.
* Function enhancements and fixes: `IFS()`, `ISREF()`, `MAP(INDEX())`, `MMULT()`, `SEQUENCE()`, `SUM(IF())`, `TEXT()`.
* Performance enhancements and bug fixes.

## `1.25.0` 2025-12-02

* Significant improvements to Excel [Data Table](https://support.microsoft.com/en-us/office/calculate-multiple-results-by-using-a-data-table-e95e2487-6ca6-4413-ad12-77542a5ea50b) performance of Excel. Note there was a fix for inconsistent results in `1.26.0`.
* Performance enhancements and bug fixes.

## `1.24.1` 2025-11-14

* Fixed an issue with formulas where the dynamic array anchor resolved to a single cell.
* Function enhancements and fixes: `INDEX()`.
* Additional updates for `request_meta` parameters echo to `response_meta`.
* Performance enhancements and bug fixes.

## `1.24.0` 2025-10-27

* Significant improvements to `LAMBDA()` functions and associated `BY_ROW()`,  `BY_COL()` functions.
* More consistent echoing of `request_meta` parameters to the `response_meta`.
* Function enhancements and fixes:  `CONCAT()`, `ERRORTYPE()`, `FORECAST.ETS.*()`,  `IF()`, `INDEX()`, `INDIRECT()`,  `ISBLANK()`, `ISERR()`, `ISERROR()`, `ISLOGICAL`, `ISNA()`, `ISNONTEXT()`, `ISNUMBER()`, `ISTEXT()`, `LEN()`, `RANK()`, `TRIM()`.
* Performance enhancements and bug fixes.

## `1.23.2` 2025-09-30

* This version of `Neuron` has debug and tracing features disabled.
* Optimizations to `IFS()` function for lazy evaluation.
* `CHOOSECOLS()` and `CHOOSEROWS()` will only evaluate the chosen cells.
* Improved `Xcall` performance for tables with many blanks.
* Improved standardization of supported `Xcall` request meta to those from [Execute API (v4)](/spark-apis/execute-api/execute-api-v4.md).
* New warning if Excel has iterative calculations enabled.
* Simplified JSON parsing of improper requests to improve performance.
* Improper `xBuildParams` will lead to an error on compile.
* Function enhancements and fixes: `COUNTA()`, `ISEVEN()`, `ISODD()`, `SLOPE()`, `TEXT()`, `VALUETOTEXT()` .
* Performance enhancements and bug fixes.

## `1.23.1` 2025-07-23

* Fixed `Xrichoutput()` bug where the last row was incorrectly identified.
* Further performance improvements for `LAMBDA()` related functions.
* Reduce some redundant `Xcall` s when there are multiple `Xcall`s within `1` Spark service.
* `Xcall` will correctly return a table output if it is the last entry in the output template and it does not have defined rows and columns.
* `Xcall` now supports additional meta in alignment with [Execute API (v4)](/spark-apis/execute-api/execute-api-v4.md).
* When direct addressing is used, extra `""` rows will not be trimmed. This helps to address some issues relating to `Xcall`.
* Align `Xcall` errors relating to subservices.
* Performance enhancements and bug fixes.

## `1.23.0` 2025-06-23

* Fix an issue with `Xcall` where a single input with an array led to an error.
* Significant performance improvements for formulas that include `LAMBDA` and  `LET` .
* Function enhancements and fixes: `ISNUMBER()`, `NUMBERVALUE()`, `TEXT()`.
* Performance enhancements and bug fixes.

## `1.22.8` 2025-04-27

* Addressed `TEXT()` function issues when constructing Indian number format.
* Function enhancements and fixes: `MAKEARRAY()`, `TEXT()`, `VLOOKUP()`.
* Performance enhancements and bug fixes.

## `1.22.7` 2025-03-24

* Addressed an issue where `#SPILL!` error was not always raised.
* Function enhancements and fixes: `BETA.DIST()`, `BETA.INV()`, `BETADIST()`, `BETAINV()`, `ISOMITTED()`, `LOGINV()`, `LOGNORM.DIST()`, `LOGNORM.INV()`, `LOGNORMDIST()`, `NORM.DIST()`, `NORM.INV()`, `NORMDIST()`, `NORMINV()`, `NORMSDIST()`, `NORMINV()`, `TAKE()`, `TEXT()`, `VALUE()`, `YEARFRAC()`.
* Performance enhancements and bug fixes.

## `1.22.6` 2024-12-20

* Address an issue for `Xcall` when the service name is not provided as a default value.
* `DEBUG` build profiler can now return performance at a worksheet, row, or column level instead of just at a cell level.
* `Xsolve` `debug_solve` `BRENT` algorithm improve logging.
* Function enhancements and fixes: `AVERAGE()`.
* Performance enhancements and bug fixes.

## `1.22.5` 2024-12-16

* `Xsolve` `debug_solve` will provide more verbose information about the solve parameters and the time to perform each iteration.
* `Xsolve` `BRENT`algorithm reduce a redundant `f(0)`evaluation in the solve.
* `Xsolve` `BRENT`algorithm introduce parameters for lower and upper bounds.
* Performance enhancements and bug fixes.

## `1.22.4` 2024-11-18

* Address an issue in overlapping input detection.
* Address an issue with `Xcall` within filled down formulas.
* Function enhancements and fixes: `IFS()`.
* Performance enhancements and bug fixes.

## `1.22.3` 2024-10-30

* Compiled `Xcall` returns validation errors if they exist in the downstream service.
* `/batch` reduce extraneous `[]` results when using `data.summary`.
* Adjust treatment of constant integers to reduce overflow scenarios.
* Function enhancements and fixes: `MAX()`, `SUMIF()`, `VLOOKUP()`.
* Performance enhancements and bug fixes.

## `1.22.2` 2024-09-26

* `Xrichoutput` to return fields in "data" order in stead of alphabetical order.
* Function enhancements and fixes: `CONCATENATE()`, `SUM()`, `VLOOKUP()`.
* Performance enhancements and bug fixes.

## `1.22.1` 2024-08-20

* Data Validation lists that contain different data types are better supported.
* Execution warning will be raised if an Excel file has overlapping `Xinput`s.
* Fixed an issue with `MATCH()` on `SPARK_XCALL()` result.
* `Xrichoutput` fix where non-`Xrichoutput` headings were duplicated into a separate object.
* The compiled Wasm ZIPs include the `{service}_CompilerLog.json` file to support the [Hybrid Runner environment variables reference](/hybrid-runner/setup-the-hybrid-runner/hybrid-runner-environment-variables-reference.md#neuron-version-compatibility) feature.
* Function enhancements and fixes: `COUNTIFS()`, `SUMIFS()`, `TEXT()`.
* Performance enhancements and bug fixes.

## `1.21.0` 2024-07-12

* Optimizations to `IF()` function for lazy evaluation.
* Neuron `Xcontroller` sheet implementation.
*  `Xcall` formula parsing improvements.
* Function enhancements and fixes: `DATE()`, `LET()`.
* Performance enhancements and bug fixes.

## `1.20.0` 2024-06-17

* `COUNTIF()`, `COUNTIFS()`, `SUMIF()`, `SUMIFS()` performance improvements and enhancements to relational operators such as `>=`.
* Data Validation fixes for explicit List validation where leadings spaces were not trimmed.
* `Xrichoutput` additional changes for more consistent handling of `""` vs `null` .
* Implement functions: `HYPERLINK()`.
* Function enhancements and fixes: `PERCENTRANK_EXCL()`, `TAKE()`, `TEXT()`.
* Performance enhancements and bug fixes.

## `1.19.2` 2024-06-04

* Restore fixes for Data Validation that were reverted in `1.19.0`.
* Improvement to logic relating to `@` implicit intersection operator when used upon an array.
* Function enhancements and fixes: `CHOOSE()`, `TEXT()`.

## `1.19.1` 2024-05-27

* Fixed bug relating to `@` implicit intersection operator when used upon an array.

## `1.19.0` 2024-05-20

* `INFO` message is raised when a Spark service includes `SPARK_XCALL()` to clarify that the self-test cannot be performed during the compilation.
* Remove upload warnings related to `1xN>1` ranges.
* Data Validation fixes for Thai characters.
* Function enhancements and fixes:  `COUNTIFS()`, `RATE()`, `REPLACEB()`, `SUMIFS()`, `TEXT()`.
* Performance enhancements and bug fixes.

## `1.18.1` 2024-05-02

* Fixes a calculation issue if a single cell formula is copied to another part of the worksheet or workbook that is not immediately to the left, right or above or below the copied cell. If this occurs, you may see mismatches between Spark and Excel in the compilation self-test performed at the end of each service upload.
* `Xrichoutput` stability enhancements and fixes. Better treatment of `""` as `""` and empty as `null`.
* Performance enhancements and bug fixes.

## `1.18.0` 2024-04-15

{% hint style="warning" %}
We advise to use Neuron `1.18.1` or newer that eliminates errors relating to copied formulas.
{% endhint %}

* Support for `CSDEV`, `CSSTG`, `CS` namespacing for Coherent Assistant custom functions.
* API warning on missing inputs is no longer case sensitive.
* Data Validation will properly consider spaces in text strings.
* Warnings will be raised when invalid ranges are defined in a Named Range and the address will swapped when used with Neuron, e.g. `Z2:A5` will be treated as `A2:Z5`.
* Improved handling of dynamic range addresses such as `A1#`.
* Files containing `Xreport` will have the automatic direct addressing enabled but restricted to `#SparkFormulaValueVector` only. See [Execute API (v3)](/spark-apis/execute-api/execute-api-v3.md) for more information.
* Function enhancements and fixes: `MIDB()`, `REPLACEB()`, `SPARK_FILTERJSON()`, `TEXT()`.
* `PHONETIC()` function is no longer supported.
* Performance enhancements and bug fixes.

## `1.17.0` 2024-03-18

{% hint style="warning" %}
We advise to use Neuron `1.18.1` or newer that eliminates errors relating to copied formulas.
{% endhint %}

* Spark will indicate which input had a default value applied in the warning messages.
* Messages raised when the request dimensions exceed the size defined in Excel has been changed from errors to warnings.
* Function enhancements and fixes: `INDIRECT()`, `MIN()`, `OFFSET()`, `SUMIFS()`, `TEXT()`.
* Direct Addressing of outputs is now managed using `xBuildParams` . See [Execute API (v3)](/spark-apis/execute-api/execute-api-v3.md) for more information. It will automatically be enabled when `Xreport` is present in the file.
* Performance enhancements and bug fixes.

## `1.16.0` 2024-02-26

{% hint style="warning" %}
We advise to use Neuron `1.18.1` or newer that eliminates errors relating to copied formulas.
{% endhint %}

* Very small numbers should return as a decimal rather than scientific notation.
* Batch API chunk support for `AVERAGE()`, `COUNT()`, `COUNTA()`, `PRODUCT()` functions.
* Dates as table headers should calculate correctly.
* Support for table references in `INDIRECT()`.
* `Xcall` default behavior to accept a single vector as an array of `Xcall` rather than a vector of inputs.
* `Xrichoutput` enhancements.
* Function enhancements and fixes: `LAMBDA()`.
* Performance enhancements and bug fixes.

## `1.15.2` 2024-02-15

* Aggregation functions will consider a missing parameter as `0`, e.g. `MAX(,-1)=0`.
* Function enhancements and fixes: `IPMT()`, `PPMT()`, `RRI()`. 

## `1.15.1` 2023-01-29

* Execute API `request_meta` `debug_solve` returns the object `outputs.additional_outputs.solve_log` which contains the iterations for each [Solve functions (Xsolve)](/build-spark-services/other-mapping-options/solve-functions.md).
* Warning will be raised if for a table input the API request headings do not match the `Xinput` definition.
* When tables are provided in the default "records" format, e.g. `[{"A":1,"B":2},{"A":3,"B":4}]` any missing columns will now take the default value. This is a change from the previous behavior. Note that this was already the behavior for JSON array format tables, e.g. `[["A","B"],[1,2],[3,4]]`.
* Further `Xrichoutput` enhancements.
* Batch API chunk support for `MIN()`, `MAX()`, `SUM()`.
* Implement functions: `CONVERT()`, `ISFORMULA()`.
* Function enhancements and fixes: `DATEVALUE()`, `LAMBDA()`, `MAKEARRAY()`. `NUMBERVALUE()`, `RIGHTB()`, `VALUE()`.
* Performance enhancements and bug fixes.

## `1.14.0` 2023-12-18

* Support for input and output merged columns.
* Support for the extended batch input format.
* Improved procesing of the JSON inputs.
* `Xsolve` algorithm `BRENT` algorithm adjust initial guess if first iteration is an error.
* Address an issue with the incorrect `Xsolve` reported solve target value.
* Enhancements in managing the calculation order of Data Tables.
* Address fixes to nested Data Tables calculations.
* Calculation time adjustments for fast services.
* Improvements in formula group support that improve the compilation time and reduction in source code.
* `Xrichoutput` enhancements.
* `INFO` message if a workbook contains both Data Tables and `Xsolve`.
* `Xcall` fix for *error template includes duplicate output*.
* `Xcall` support for horizontal and vector column tables.
* Support for future Coherent Assistant custom function namespaces.
* Implement functions: `AREAS()`, `BITAND()`, `BITLSHIFT()`, `BITOR()`, `BITRSHIFT()`, `BITXOR()`, `DELTA()`, `DOLLAR()`, `DSTDEVP()`, `NCODEURL()`, `GESTEP()`, `HEX2BIN()`, `HEX2OCT()`, `IMDIV()`, `IMPRODUCT()`, `SHEET()`, `SHEETS()`, `T.INV()`, `VDB()`, `YIELD()`.
* Function enhancements and fixes: `NUMBERVALUE()`.
* Performance enhancements and bug fixes.

## `1.13.1` 2023-11-20

* Includes the changes from `1.12.2` and `1.13.0`.
* Improved handling of formulas that reference an entire column improving compilation times for such models.
* Better support for array expressions in formulas `{1;1;..;1}`.
* Performance enhancements and bug fixes.

## `1.12.3` 2023-11-17

* `Xsolve` algorithm `BRENT` algorithm adjust initial guess if first iteration is an error.
* Improvements in formula group support that improve the compilation time and reduction in source code.

## `1.12.2` 2023-11-10

* Support multiple sets of array inputs in array input `Xcall`.
* For `Xsolve`, introduce the fields `Solve Started` and `Solve Successful` in order to better understand the progress of a solve.
* `MAKEARRAY()` support for `LAMBDA()`.
* Better handling of workbooks with Data Tables and `Xsolve`. An `INFO` message will be noted by the compiler.
* Resolve the `Xsolve` issues introduced in `1.12.1`.
* Performance enhancements and bug fixes.

## `1.13.0` 2023-10-25

* Friendly Neuron compiler and self-test messages.
* Fixed compilation errors for Excel files with dynamic arrays.
* `Xcall` now supports multiple sets of array inputs.
* `Xcall` array outputs include any corresponding validation errors in the downstream service.
* Partial support for nested Data Tables.
* Circular reference fix.
* Fixed comparisons between empty cells and empty string.
* Fix `Xrichoutput` errors when used with `Xprecalc`.
* Fix `Xvalidate` message not showing.
* Fix for data validation errors for time inputs.
* Additional `Xsolve` field `Solve started`, `Solve successful`.
* This release introduced a bug where in certain rare cases where models containing interdependencies between multiple Data Tables and `Xsolve`s may produce unstable results.
* Implement functions: `BETAINV()`, `BYROW()`, `CHISQ.INV.RT()`, `CHITEST()`, `HEX2DEC()`, `IMABS()`, `IMAGINARY()`, `IMARGUMENT()`, `IMCONJUGATE()`,  `IMCOSH()`, `IMCOT()`, `IMCSC()`, `IMCSCH()`, `IMPOWER()`,  `IMSEC()`, `IMSINH()`, `ISOMITTED()`, `NETWORKDAYS.INTL()`, `NORMS.S.INV()`, `REDUCE()`, `TDIST()`, `TTEST()`, `UNICHAR()`, `UNICODE()`. 
* Function enhancements and fixes: `CELL()`, `FORECAST()`, `FORECAST.LINEAR()`, `GAMMA()`, `INFO()`, `LOGNORM.DIST()`, `SORT()`, `XLOOKUP()`.
* Performance enhancements and bug fixes.

## `1.12.1` 2023-10-04

* This release introduced a bug which may give incorrect values for `Xsolve`.
* Function enhancements and fixes: `ASC()`, `INDIRECT()`.
* Performance enhancements and bug fixes.

## `1.12.0` 2023-09-25

* Fix for `Xcall` not running when referenced indirectly in a compiled service.
* Fix for `Xcall` not working with array/spill functions in input location/output template.
* Fix for `Xcall` subservice not populating output.
* Fix for recursive `LAMBDA()` not terminating.
* Fix for Named ranges containing `#REF` errors were affecting execution.
* Fix for `Xcall` not working consistently for single output requested in array Xcall.
* Fix for `Xcall` without an output template.
* Fix Wasm package verification runner timed out error on `Xcall` array spec.
* Fix the calculation time bug from the previous release.
* Implement functions: `ASC()`, `BYCOL()`, `DBCS()`, `DROP()`, `ERF`, `ERF.PRECISE()`, `ERFC()`, `ERROR.TYPE()`,  `FORECAST.ETS.STAT`, `FVSCHEDULE()`, `IMCOS()`, `IMEXP()`, `IMLN()`, `IMLOG10()`, `IMLOG2()`, `IMREAL()`, `IMSECH()`, `IMSIN()`, `IMSQRT()`, `IMSUB()`, `IMSUM()`, `IMTAN()`, `JIS()`, `MODE.MULT()`, `OCT2BIN()`, `OCT2DEC()`, `OCT2HEX()`, `PERCENTRANK.EXC()`, `PROB()`,`RANDARRAY()`, `SHEET()`, `SKEW.P()`, `Z.TEST()`.
* Function enhancements and fixes: `DATEDIF()`, `FIXED()`, `MATCH()`.
* Performance enhancements and bug fixes.

## `1.11.0` 2023-08-29

* Fix warning message for input column dimensions exceeding the defined input range in Excel.
* `XCSVinput` performance enhancements.
* Implement functions: `EXPAND()`, `F.INV.RT()`, `FILTERXML()`, `FINV()`, `SUMIFS()`, `TAKE()`,  `TINV()`, `TOCOL()`, `TOROW()`, `TRIMMEAN()`, `WRAPCOLS()`, `WRAPROWS()`.
* Function enhancements and fixes: `COUNT()`, `INDIRECT()`, `TEXT()`.
* Performance enhancements and bug fixes.
* Bug introduced relating to calculation times.

## `1.10.1` 2023-08-01

* Circular references use the iterative calculation properties saved in the workbook.
* Warning is raised if `""` is submitted as a single input.
* Warning is raised if the Excel output contains an error.
* Data validations fixed for large numbers.
* `Xcall` support for `service_category` for input parameters.
* Array input and output spill fixes.
* Implement functions: `CHISQ.DIST()`, `CHISQ.TEST()`.
* Function enhancements and fixes: `INDEX()`, `INDIRECT()`, `SUMPRODUCT()`, `TEXT()`.
* Performance enhancements and bug fixes.

## `1.10.0` 2023-07-24

* `Xrichoutput` fixes (noted: led to some subsequent issues reported by clients).
* `Xcall` support for an array of names.
* Improved handling of dynamic range vector output for `SEQUENCE()`, `SORT()`, `XLOOKUP()`, etc.
* `Xcall` no longer requires output template.
* `Xcall` no longer requires table headers in addition to the output name for table outputs.
* Asynchronous and synchronous array JSON format support.
* `Xcall` Ignore trailing empty cells in 1xN output template.
* Implement functions: `BETADIST`,  `CHISQ.INV`, `CHITEST()`, `CHOOSECOLS()`,  `CRITBINOM()`, `HSTACK()`, `INFO()`, `T.DIST()`, `XOR()`.
* Function enhancements and fixes: `AGGREGATE()`, `CHOOSE()`, `INDIRECT()`, `SUBTOTAL()`, `TEXT()`, `VALUE()`, `VLOOKUP()`.
* Performance enhancements and bug fixes.

## `1.9.0` 2023-06-26

This release coincided with the feature to [Manage tenant settings](/tenant-administration/manage-tenant-settings.md#choose-the-neuron-compiler-for-newly-uploaded-services).

* `Xrichoutput` fixes (noted: led to some subsequent issues reported by clients).
* Handling of `/` and `~` in column names.
* `Xcall` support for subservices in `request_meta`.
* Resolved external reference leading to error message.
* Fixed `#REF` error showing when using `Xcall` and upstream service has `Xoutput`.
* Fixed a bug where `XCSVoutput` crashed compiler.
* `Xcall` `1xN` output template handling improvements.
* Implement functions: `FORECAST.LINEAR()`, `MAP()`.
* Function enhancements and fixes: `DATEVALUE()`, `NUMBERVALUE()`, `PRODUCT()`, `SUBTOTAL()`, `SUMPRODUCT()`, `TIMEVALUE()`, `VALUE()`,  `XLOOKUP()`.
* Performance enhancements and bug fixes.

## `1.8.2` 2023-06-16

Inclusive of changes in `1.7.0`, `1.8.0`, and `1.8.1`.

* Improvement for compiler messages.
* Improved `Xcall` caching mechanic.
* Circular reference support.
* `Xcall` support for `1xN` output template.
* Fixed bug where 2 `@` in a function led to incorrect output.
* Fixed requested output direct reference not including first row.
* CSE formula fixes.
* `Xcall` support for `request_meta`.
* Improvements to direct addressing outputs.
* In merged header tables, handle dotted column names as subheadings.
* Improve the handling of JSON data.
* Self-test includes subservice outputs.
* `1xN` vector remove validation for header duplication.
* Function enhancements and fixes: `NUMBERVALUE()`.
* New warnings relating to ill-formed `LAMBDA()` function using `MAKEARRAY()`.
* `Xcall` `1xN` output template bug fixes relating to spill cells.
* Implement functions: `CHOOSECOLS()`, `CHOOSEROWS()`.
* Function enhancements and fixes: `ABS()`, `SUMIF()`, `SUMPRODUCT()`.
* Performance enhancements and bug fixes.

## `1.6.2` 2023-05-05

Inclusive of changes in `1.6.0` and `1.6.1`.

* Resolve issues relating to `@` references.
* `OutofMemoryException` fixes.
* Resolve errors relating to missing heading columns.
* Better handling of date and time inputs.
* Add support of anchor array for named ranges.
* Implement functions: `TEXTBEFORE()`.
* Function enhancements and fixes: `COUNTIF()`, `FILTER()`, `FLOAT()`, `GET.WORKBOOK()`, `IFERROR()`, `INDEX()`, `INFO()`, `INT()`,  `MID()`, `TEXT()`.
* Performance enhancements and bug fixes.

## `1.5.0` 2023-04-14

* Implementation of warning when the request will rely upon default values.
* Circular formula detection.
* Implementation of the `SPARK_XCALL()` UDF.
* `Xcall` `1xN` table function support.
* Outputs linked to an input without proper headings were incorrect.
* Correct the condition where the self-test incorrectly shows a mismatch when text format is applied to a number.
* API bug for particular `Xinput` names not properly parsed.
* `Xcall` subheader table support for input and output templates.I
* Increase the maximum memory allocation for the compiler.
* Optimize `XLOOKUP` by presorting the array and perform binary search.
* Performance enhancements and bug fixes.

## `1.4.0` 2023-03-27

This is the first version of Neuron that is made available via the [Tenant administration](/tenant-administration.md) [Manage tenant settings](/tenant-administration/manage-tenant-settings.md#choose-the-neuron-compiler-for-newly-uploaded-services) functionality.

* Fix for `Model execution error` with Metadata subservice.
* Array inputs and outputs with subheaders.
* Warnings will appear if the request input exceeds the dimensions defined in Excel.
* Fixed handling of nulls in external references.
* Improved stack overflow handling.
* Improved handling of relative area references.
* `Xjinput` value issue fix to configure table of inputs.
* Improved date and time handling in API response.
* Implement functions: `PHONETIC()`, `SUBTOTAL()`.
* Function enhancements and fixes:  `DATEDIF()`, `DATEVALUE()`, `IF()`, `IFNA()`, `LEFT()` and `YEARFRAC()`.
* Performance enhancements and bug fixes.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.coherent.global/build-spark-services/neuron/neuron-release-history.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.