Fastest
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →https://eureka.patsnap.com/resources/blog/rd-blog/language-model-code-generation-syntax-aware-decoding-patent-landscape-patent-landscape/ · Patsnap · data cut-off 2026-08-31 · downloaded from the live page
Patent Landscape · Artificial Intelligence & Machine LearningLanguage Model Code Generation Syntax-Aware Decoding Patents
A patent landscape review of language model code generation with syntax-aware decoding: 94 records, filing trends from 2015, assignee concentration, IPC composition, and white space across 81 ranked companies through 202
94
Published Records
29%
Top-5 Share of All Records
+950%
Filing Growth 2021→2024
CN
Leading Jurisdiction
Filing growth = 2021 (2 records) → 2024 (21); 2024 is the last year we treat as complete. Top-5 share = the 5 largest assignees ÷ all 94 records in scope (CR5), not the ranked leaders only.
Published byPatsnap Research··6 min readSourced from Patsnap Eureka
Overview
What this landscape covers
This landscape tracks patent filings at the intersection of large language models, code generation or program synthesis, and syntax awareness — inventions that constrain, validate or correct the syntactic structure of machine-generated code using grammars or abstract syntax trees. The dataset spans 94 published records filed between 2015 and the 2026 cut-off, ranked across 81 assignees. Filing activity was negligible before 2021 and then accelerated sharply, which places most of the substantive claim activity in the last five years.
Because publication lags actual filing by roughly 18 months, the most recent one to two years in any trend line will always look lighter than filing activity actually was — that effect is visible in this dataset's 2025-2026 counts and should not be read as a slowdown.
Filing volume and IPC composition, 2015-2026
Source: Patsnap Eureka. Assignee ranking and totals. Derived from a Patsnap search on Language Model Code Generation Syntax-Aware Decoding Patent Landscape covering 2015–2026, data cut-off 2026-08-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Filing Data
Filing trends and technology composition
The dataset spans 94 published records from 2015 through the 2026 cut-off, tracked across 81 ranked assignees and eight IPC subclasses.
Filing activity climbed sharply from 2021 onward
Filings moved from 2 in 2021 to 21 in 2024, a +950% rise, and continued to 2025's peak of 38 before the partial 2026 count of 18. Because publication lags filing by roughly 18 months, 2025 and 2026 are undercounts, not a plateau or a decline.
Publication lags filing by roughly 18 months, so 2025 onwards are still filling in. Growth rates on this page therefore end at 2024; running them to the last bar would understate the field.
G06F and G06N carry almost all of the claim weight
G06F (electric digital data processing) appears in 92.6% of the 94 records and G06N (AI computing models) in 53.2%, confirming this is claimed primarily as a data-processing and model-architecture invention. The remaining six subclasses — covering imaging, robotics, business processing, video recognition and education — each sit at 3.2% or below, marking where syntax-aware code generation has barely been claimed outside its core software domain.
Shares are the percentage of the 94 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Source: Patsnap Eureka. Filing trend and technology composition. Derived from a Patsnap search on Language Model Code Generation Syntax-Aware Decoding Patent Landscape covering 2015–2026, data cut-off 2026-08-31. Counts reflect published records only and shift as new filings publish.
Go deeper on Language Model Code Generation Syntax-Aware Decoding Patent Landscape with Eureka
This page is one run against one query. Ask Eureka your own question about language model code generation syntax-aware decoding patent landscape and every answer comes back with the patent numbers behind it.
Try EurekaRepresentative Filing
A representative claim in this space
Representative Record
Semantic Interpreter for Natural Language Commanding in Applications Via Program Synthesis
MICROSOFT TECHNOLOGY LICENSING, LLC
A data processing system implements receiving a natural language query from a client device, the query including a textual description of one or more actions to be performed using an application on the client device, generating a prompt for a large language model (LLM) based on the natural language query; providing the prompt to the LLM to obtain program code, which when executed, causes the application to perform the one or more actions, the first program code being implemented in a domain specific language (DSL) designed to facilitate program synthesis via the LLM; analyzing a syntax of the first program code to identify syntax errors in the first program code; correcting the syntax errors.Filed by Microsoft Technology Licensing, LLC, published 2024-11-14. Shown here as representative of the DSL-constrained, correction-pass approach that recurs across the dataset.
Source: Patsnap Eureka. Citation counts and representative records. Derived from a Patsnap search on Language Model Code Generation Syntax-Aware Decoding Patent Landscape covering 2015–2026, data cut-off 2026-08-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Run it yourself
Put your own technology through the same analysis
Where to run it
For builders
MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →Signals
What the filing data signals for strategy
Four figures from the dataset matter more than the rest for deciding where to file, watch, or design around existing claims.
Growth
+950%
2021 → 2024 filings
The field is still being actively staked
Filings rose from 2 in 2021 to 21 in 2024, and the raw count kept climbing to a peak of 38 in 2025. This is not a maturing, slowing field — it is one where claim territory is still being drawn, which raises the cost of waiting to file.
Based on the 2021-2024 complete-year window.
Concentration
28.7%
of 94 records held by top 5
No single filer controls the field
The leading assignee holds 13 records, but the top 5 combined account for only 28.7% of the 94 records in scope, and the top 10 reach 39.4%. Beyond the leader, ownership thins fast — fifth and tenth place each hold just 2 records.
Ranking covers 81 companies total.
Composition
92.6%
of records tagged G06F
Claims sit on core data processing, not niche AI
G06F (electric digital data processing) appears in 92.6% of the 94 records and G06N (AI models) in 53.2%. The overlap confirms most inventions combine a data-processing mechanism with an AI-model element rather than claiming either alone.
Class shares sum above 100% because records carry multiple IPC codes.
Geography
35 / 34
China vs US filings
Clearance work starts in two jurisdictions
China (35) and the United States (34) account for the bulk of filing activity, with India (11), EPO (6) and WIPO (6) well behind and Japan at just 1. Any freedom-to-operate check should prioritise these two offices first.
Receiving-office counts from the tracked dataset.
Eureka AI Agent
Looking for what nobody has claimed yet?
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to language model code generation syntax-aware decoding patent landscape, with the prior art for and against each one.
Source: Patsnap Eureka. Derived observations on the same dataset. Derived from a Patsnap search on Language Model Code Generation Syntax-Aware Decoding Patent Landscape covering 2015–2026, data cut-off 2026-08-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Next Steps
Where to take this next
The figures above set the boundaries of the field. Turning them into a filing or freedom-to-operate decision means going record by record against your own architecture.
Check a specific implementation against the leader's claims
If your pipeline combines DSL-constrained generation with post-hoc syntax correction, the representative Microsoft record is the closest prior art to test against before drafting new claims.
Compare claim scope in EurekaScope the under-claimed branches before a competitor does
Robotics, business-process automation, and education applications of syntax-aware code generation each sit at 3.2% of records or below — worth a deeper look while the space is still open.
Explore white space in EurekaTrack the leading filers as the field keeps growing
With filings up +950% from 2021 to 2024 and no single filer holding more than a modest share, new entrants are still shaping the field — ongoing monitoring matters more here than in a settled art.
Set up monitoring in EurekaSource: Patsnap Eureka. Forward-looking reading of the same dataset. Derived from a Patsnap search on Language Model Code Generation Syntax-Aware Decoding Patent Landscape covering 2015–2026, data cut-off 2026-08-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
FAQ
Frequently asked questions
The tracked dataset contains 94 published records filed between 2015 and the 2026 cut-off, drawn from a search across code generation, program synthesis, and syntax/grammar/AST terms combined with language model or LLM terminology. This counts patent families in the assignee ranking, which is the fairer unit than raw document counts because it neutralises duplicate continuation filings and multi-jurisdiction copies of the same invention. Filing activity was minimal before 2021 and accelerated sharply afterward, so the great majority of these records are recent.
Filing is moderately concentrated: the leading assignee holds 13 of the 94 records, and the top 5 combined account for 28.7% of all records in scope, rising to 39.4% for the top 10. Beyond the leader, the field thins quickly — fifth place and tenth place each hold only 2 records — which points to a long tail of single- or double-filing entrants rather than a small, closed group of dominant players. The ranked list covers 81 companies total, so it is a broad but not evenly distributed field.
The overwhelming majority sit in G06F (electric digital data processing, 92.6% of the 94 records) and G06N (AI computing models, 53.2%), confirming the field is claimed mainly as core software and model-architecture invention. Smaller pockets extend into image data processing, robotics manipulators, business-process automation, video recognition and educational tooling, but each of these accounts for only 1.1% to 3.2% of records. That gap is a signal of under-claimed territory rather than a lack of commercial relevance in those adjacent domains.
China and the United States lead as receiving offices, with 35 and 34 filings respectively out of the tracked dataset, followed by India at 11. Europe (EPO) and WIPO (PCT) each show 6 filings, and Japan trails with just 1. This split suggests that clearance and freedom-to-operate work should prioritise China and the US first, since together they account for the bulk of the filed art, with India as a smaller but non-trivial third market.
Filing grew from 2 records in 2021 to 21 in 2024, a documented +950% increase, and the raw count continued up to a peak of 38 in 2025. It would be a mistake to read the lower 2025 and partial-2026 figures as a slowdown: publication typically lags actual filing by around 18 months, so the most recent one to two years are always undercounted in any patent dataset at the point of measurement. The 2021-2024 window is the most reliable evidence of trend, and it shows sustained, fast growth rather than deceleration.
Answers are grounded in the same dataset. Derived from a Patsnap search on Language Model Code Generation Syntax-Aware Decoding Patent Landscape covering 2015–2026, data cut-off 2026-08-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Research Language Model Code Generation Syntax-Aware Decoding Patent Landscape in depth with Eureka
Go past this page: query the whole language model code generation syntax-aware decoding patent landscape corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
Disclaimer. This page is generated from Patsnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. Patsnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company's registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.