Prosecution Insights
Last updated: April 19, 2026
Application No. 18/211,967

LARGE LANGUAGE MODEL (LLM) QUANTIZATION

Non-Final OA §101
Filed
Jun 20, 2023
Examiner
JOHNSON, AMY COHEN
Art Unit
2400
Tech Center
2400 — Computer Networks
Assignee
Google LLC
OA Round
1 (Non-Final)
57%
Grant Probability
Moderate
1-2
OA Rounds
2y 7m
To Grant
80%
With Interview

Examiner Intelligence

Grants 57% of resolved cases
57%
Career Allow Rate
284 granted / 499 resolved
-1.1% vs TC avg
Strong +23% interview lift
Without
With
+22.9%
Interview Lift
resolved cases with interview
Typical timeline
2y 7m
Avg Prosecution
342 currently pending
Career history
841
Total Applications
across all art units

Statute-Specific Performance

§101
3.9%
-36.1% vs TC avg
§103
55.7%
+15.7% vs TC avg
§102
21.4%
-18.6% vs TC avg
§112
10.9%
-29.1% vs TC avg
Black line = Tech Center average estimate • Based on career data from 499 resolved cases

Office Action

§101
Detailed Action 1. This office Action is in response to the application filed on 06/20/23. Claims 1-20 are pending and are examined. Notice of Pre-AIA or AIA Status 2. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA . Information Disclosure Statements 3. The information disclosure statements (IDS) submitted on 06/20/23 and 08/20/23 are being considered by the examiner. Claim Rejections - 35 USC § 101 35 USC § 101 reads as follows: Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvements thereof, may obtain a patent therefor, subject to the conditions and requirements of this title. 4. Claim 1. is rejected under 35 USC § 101 because the claimed invention is directed to non-statutory subject matter. The claim does not fall within at least one of the four categories of patent eligible subject matter because the claimed invention is directed to software per se. Specifically, "processors" as entities unto themselves are considered to be software. Correction of this issue can be made by substituting "computer processors" for the term "processors". 5. Dependent claim 2. - 7. Are claims dependent from Claim 1. and are rejected under 35 USC § 101 for similar reason as those noted above since such claims do not cure the rejection of independent Claim 1. 6. Claim 1 is rejected under 35 U.S.C. § 101 as being directed to a judicial exception without significantly more. Step 1: Statutory Category Analysis Claim 1 recites a “method,” which falls within the statutory category of a process under 35 U.S.C. § 101. Step 2A, Prong One: Judicial Exception Identified the claim as directed to an abstract idea, specifically falling within the categories of Mathematical Concepts and Mental Processes as identified in the 2019 Revised Patent Subject Matter Eligibility Guidance (84 Fed. Reg. 50). The core of the claim involves: A. Mathematical calculations — “calculating an optimal clipping range” involves mathematical relationships and formulas B. Mathematical transformations — “clipping one or more weights” and “mapping weights from continuous values to discrete values” are mathematical operations that transform numerical data C. Data processing — the manipulation of numerical weight values using mathematical algorithms These operations constitute mathematical concepts under MPEP § 2106.04(a)(2). The claim describes a mathematical algorithm for reducing the precision of numerical data (model weights), which could conceptually be performed by a person with sufficient time and computational tools. See Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350 (Fed. Cir. 2016) (holding that collecting, analyzing, and manipulating data constitutes abstract ideas). Step 2A, Prong Two: Practical Application Analysis The claim does NOT integrate the abstract idea into a practical application because: A. No Improvement to Computer Functionality. The claim does not improve the functioning of a computer or other technology. Rather, it uses the computer as a tool to execute mathematical calculations. The claim does not specify: How the processors are configured differently Any technical improvement to computer memory, processing speed, or efficiency A specific technical solution to a technical problem While model quantization may improve downstream inference efficiency, the claim itself does not recite these improvements or tie the mathematical process to any specific technological advancement. The claimed method is directed to the mathematical process itself, not to an improved computer system (Enfish, LLC v. Microsoft Corp., 822 F.3d 1327; Fed. Cir. 2016; finding eligibility where claims were directed to specific improvement in computer capabilities). B. Generic Computer Implementation. The recitation of “one or more processors” is merely a generic invocation of a computer to perform the mathematical steps. The claim does not require any particular processor architecture, specialized hardware, or non-conventional configuration. See Alice, 573 U.S. at 223 (noting that generic computer implementation is insufficient). C. Data Input/Output The additional elements, “obtaining a trained large language model” and “providing the quantized LLM for downstream processing”—are insignificant extra-solution activity: Obtaining is merely data gathering (Classen Immunotherapies, Inc. v. Biogen IDEC, 659 F.3d 1057 (Fed. Cir. 2011)) Providing for downstream processing is a field-of-use limitation and insignificant post-solution activity that does not impose meaningful limits (Parker v. Flook, 437 U.S. 584 (1978)) Therefore, the abstract idea is not integrated into a practical application. Step 2B: Significantly More Analysis Because the claim is directed to an abstract idea under Step 2A, the analysis proceeds to Step 2B to determine whether additional elements, individually or as an ordered combination, provide “significantly more.” The additional elements are: A. “implemented using one or more processors” B. “obtaining a trained large language model” C. “providing the quantized LLM for downstream processing” Analysis: A. Generic Computing Element: The “one or more processors” limitation amounts to no more than generally linking the abstract idea to a particular technological environment or field of use. This is insufficient under Alice. The claim does not recite how the processors are specifically programmed or structured to achieve the result. B. Routine Data Gathering: Obtaining input data (the trained LLM) is well-understood, routine, conventional activity in the field. MPEP § 2106.05(d). C. Insignificant Extra-Solution Activity: Providing output for unspecified “downstream processing” adds only a generic output step that does not meaningfully limit the claim or provide technological innovation. D. Ordered Combination: The ordered combination of these elements does not transform the claim into something significantly more than the abstract idea. The steps simply describe a conventional computer performing standard mathematical operations on data structures. There is no unconventional sequence, no particular machine or manufacture improvement, and no transformation beyond abstract mathematical manipulation. Conclusion: The additional claim elements do not amount to significantly more than the judicial exception itself. The claim amounts to nothing more than instructing a practitioner to implement the abstract idea using generic computer components performing conventional functions. Therefore, claim 1 is not directed to patent-eligible subject matter under 35 U.S.C. § 101. Relevant Case Law Supporting This Rejection: A. Alice Corp. v. CLS Bank Int’l, 573 U.S. 208 (2014) B. Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 U.S. 66 (2012) C. Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350 (Fed. Cir. 2016) D. Parker v. Flook, 437 U.S. 584 (1978) 7. Independent Claims 8. and 15. have similar limitations to that of Claim 1 and are rejected under 35 USC § 101 for similar reasons as noted above. Dependent claims 2.-7., 9.-14. and 16.-20. do not correct the related independent claim from which they depend and are in consequence rejected under 35 USC § 101 for similar reasons as noted above. 8. Claim Amendment Strategies to properly respond to the above rejections are informally provided as examples for applicant to consider in concept but must be consistent within the metes and bounds of the specification. A. Add Technical Improvements to Computer Functionality. Amend the claim to recite specific improvements to the computer’s operation: Example amendments: "wherein the quantized LLM requires at least 50% less memory storage than the trained LLM” “wherein the quantizing reduces computational latency during inference operations by a factor of at least 2x” “wherein the quantized LLM is configured for deployment on edge devices with limited memory capacity” “wherein the clipping and quantizing preserve model accuracy within 2% of the trained LLM while reducing model size” Supporting case law: Enfish, LLC v. Microsoft Corp., 822 F.3d 1327 (Fed. Cir. 2016) — claims directed to specific improvements in computer capabilities may be patent-eligible. B. Specify the Clipping Algorithm with Technical Detail. Add specific technical details about how the optimal clipping range is calculated: Example amendments: “wherein calculating the optimal clipping range comprises: analyzing a distribution of the respective plurality of weights to identify percentile thresholds that minimize quantization error while maximizing compression ratio” “wherein the optimal clipping range is calculated using a calibration dataset to minimize mean squared error between quantized and original weight values” “wherein calculating includes iteratively adjusting clipping thresholds based on layer-specific sensitivity analysis” Rationale: Specific algorithmic implementations may demonstrate a non- conventional technical solution. C. Tie to Specific Hardware or Architecture. Limit the claim to specific computing architectures. Example amendments: “wherein the one or more processors comprise tensor processing units (TPUs) configured with reduced-precision arithmetic logic units” “wherein the quantized LLM is configured for execution on processors supporting INT8 operations” “configuring hardware accelerators to process the quantized LLM using fixed-point arithmetic operations” Supporting case law: McRO, Inc. v. Bandai Namco Games Am. Inc., 837 F.3d 1299 (Fed. Cir. 2016) — specific technological implementation may establish eligibility. D. Specify Downstream Processing; Replace the generic “downstream processing” with specific technical applications: Example amendments: “deploying the quantized LLM to perform real-time natural language processing on mobile devices with memory constraints of less than 4GB” “executing inference operations using the quantized LLM at reduced power consumption compared to the trained LLM” “providing the quantized LLM to a neural network inference engine for accelerated text generation” E. Add Layer-Specific Technical Details; Include technical details about how different layers are processed. Example amendments: “wherein attention layers are clipped using a first clipping range and feed-forward layers are clipped using a second, different clipping range based on layer sensitivity analysis” “wherein early layers of the LLM are quantized to 8-bit precision and later layers are quantized to 4-bit precision based on layer importance scores” 9. Argument Strategies (Without Amendment). Argue practical ppplication Under Step 2A, Prong Two. Draft arguments showing integration into a practical application: A. The claim solves a specific technical problem: deploying large neural networks on resource-constrained devices B. The method improves computer functionality by enabling efficient model storage and execution C. The quantization maintains model performance while achieving concrete technical benefits (cite specification for evidence) D. Distinguished from Alice: This is not a fundamental economic practice or abstract mathematical concept divorced from technological application. Evidence to cite: Point to specification sections describing technical problems solved, performance benchmarks, memory savings, or hardware compatibility improvements. E. Argue Unconventional Implementation under Step 2B. Demonstrate that the combination is not routine or conventional: The layer-by-layer approach with individualized clipping ranges is not conventional in the field The ordered combination of clipping before quantization solves the technical problem of quantization error accumulation The method represents a departure from conventional uniform quantization approaches Cite expert declarations or technical literature showing the approach is non-obvious or unconventional Supporting case law: Ancora Techs., Inc. v. HTC Am., Inc., 908 F.3d 1343 (Fed. Cir. 2018) — unconventional technical solution may provide significantly more. 8. Demonstrate Technological Innovation Through Evidence. Submit a declaration under 37 C.F.R. § 1.132 with technical evidence: A. Comparative data showing improved performance metrics B. Technical explanations of how the method differs from prior approaches C. Evidence of commercial adoption or industry recognition D. Expert testimony on the technological significance 9. Specification Mining Strategies. Identify Overlooked Technical Details in Specification A. Look for: Specific algorithms or formulas for clipping calculation Performance benchmarks or experimental results Hardware configurations or deployment scenarios Technical problems solved by the invention Comparisons with prior art methods B. Amend to recite: Specific memory/performance improvements Technical clipping algorithm details Concrete downstream application C. PLUS argue: Integration into practical application … very important Unconventional technical solution Distinguished from cited case law Conclusion 10. The prior art made of record and not relied upon is considered pertinent applicant's disclosure: See IDSs 11. Any inquiry concerning this communication or earlier communications from the examiner should be directed to Joseph P. Hirl whose telephone number is (571)272-3685. The examiner can normally be reached Monday - Thursday 5:30 am to 3:30 pm. Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice. 12. If attempts to reach the examiner by telephone are unsuccessful, the examiner's Director, Amy C. Johnson can be reached on 571-272-2238. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300. Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patentcenter for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000. /JOSEPH P HIRL/Supervisory Patent Examiner, Art Unit 2435
Read full office action

Prosecution Timeline

Jun 20, 2023
Application Filed
Jan 24, 2026
Non-Final Rejection — §101 (current)

Precedent Cases

Applications granted by this same examiner with similar technology

Patent 12381794
METHOD AND SYSTEM FOR PERFORMING AD HOC DIAGNOSTICS, MAINTENANCE, PROGRAMMING, AND TESTS OF INTERNET OF THINGS DEVICES
2y 5m to grant Granted Aug 05, 2025
Patent 12381816
POLICY PLANE INTEGRATION ACROSS MULTIPLE DOMAINS
2y 5m to grant Granted Aug 05, 2025
Patent 12363582
METHOD FOR MANAGING QOS, RELAY TERMINAL, PCF NETWORK ELEMENT, SMF NETWORK ELEMENT, AND REMOTE TERMINAL
2y 5m to grant Granted Jul 15, 2025
Patent 12363588
DATA TRANSMISSION METHOD AND APPARATUS, COMPUTER READABLE MEDIUM, AND ELECTRONIC DEVICE
2y 5m to grant Granted Jul 15, 2025
Patent 12363337
CODING AND DECODING OF VIDEO CODING MODES
2y 5m to grant Granted Jul 15, 2025
Study what changed to get past this examiner. Based on 5 most recent grants.

AI Strategy Recommendation

Get an AI-powered prosecution strategy using examiner precedents, rejection analysis, and claim mapping.
Powered by AI — typically takes 5-10 seconds

Prosecution Projections

1-2
Expected OA Rounds
57%
Grant Probability
80%
With Interview (+22.9%)
2y 7m
Median Time to Grant
Low
PTA Risk
Based on 499 resolved cases by this examiner. Grant probability derived from career allow rate.

Sign in with your work email

Enter your email to receive a magic link. No password needed.

Personal email addresses (Gmail, Yahoo, etc.) are not accepted.

Free tier: 3 strategy analyses per month