Pith. sign in

REVIEW 4 major objections 7 minor 33 references

Email as the Interface to Generative AI Models: Seamless Administrative Automation

T0 review · 4 major / 7 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper argues that ordinary email can serve as the interface to an LLM-based administrative automation pipeline, and its proof of concept reports a three-to-four-fold reduction in staff time and a 64% cost saving per form.

desk verdict A plausible email-as-interface proof of concept whose headline efficiency and cost claims rest on an unmeasured five-minute supervision assumption. read the letter →

arxiv 2506.23850 v1 pith:IE5QTWKP submitted 2025-06-30 cs.HC

Andres Navarro , Carlos de Quinto , José Alberto Hernández This is my paper · ORCID
classification cs.HC
keywords GenerativeAILargeLanguageModelsAdministrativeAutomationEmailInterfacesOpticalCharacterRecognitionCostReductionWorkflowEfficiencyFormFilling
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that a familiar tool, email, can be the interface between non-technical administrative staff and a large-language-model (LLM) automation pipeline, removing the training and software barriers that usually block adoption. In the proposed design the email body acts as a natural-language instruction and attachments carry the documents, so an employee sends a request and receives a completed form back. The proof of concept fills a 29-field Spanish administrative form in about 7.8 seconds of automated processing, with the best model filling 16 fields correctly and leaving 11 blank rather than guessing. Counting roughly five minutes of human review per form, the authors estimate staff time falls by a factor of three to four and direct cost per form falls by about 64% compared with manual completion. The point of the claim is that organizations can capture value from advanced models without retraining workers or redesigning workflows.

What carries the argument

The load-bearing mechanism is the email-as-prompt pipeline: an email client watches an inbox, an OCR module converts attached documents to text, an LLM turns the email body plus that text into a fill plan, and Python automation scripts execute the plan on a PDF form. The design makes the email body the natural-language instruction channel and attachments the context channel, so no new interface or training is required. The evaluation's optimal-result row is also part of the machinery: it defines the target behavior of leaving a field blank unless the supplied context contains the information, turning hallucination avoidance into a scored property of the model.

What would settle it

Run a head-to-head pilot on the same form, having administrative staff complete it manually while the automated pipeline processes another batch under real working conditions and recording actual review time per form; if average review time exceeds about 15 minutes, the claimed factor-of-three-to-four staff-time saving no longer holds.

Watch

Extended reading notes

Core claim

The central claim, on the paper's own terms, is that an email interface plus optical character recognition (OCR) plus an LLM plus small automation scripts is a viable and cost-effective way to automate structured administrative work. Treating the email body as a prompt and attachments as context lets the system parse a request, extract text from scanned documents, decide which fields can be answered, and return a filled PDF. The authors benchmark seven LLMs on a 29-field form and find llama-4-maverick-17b-128e-instruct best at 16 correct, 2 incorrect, and 11 blank, against an optimal result of 17 filled and 12 blank when no information is provided. They report 7.8 seconds of automated OCR and LLM processing and, with 5 minutes of human supervision, total staff time per form reduced by a factor of three to four and cost reduced by 64.6%, from €4.81 to €1.70 per form. The paper frames this as democratizing automation: the gains come from interface familiarity rather than technical expertise.

Load-bearing premise

The reported time and cost savings rest on the assumption that each form needs only about five minutes of human review and that loaded staff time costs roughly €19.23 per hour, so if review runs longer or wages differ, the three-to-four-fold time reduction and 64% savings shrink.

Editorial extensions

If this is right

  • Organizations can deploy LLM-based form automation without retraining administrative staff, because the interaction is the same email workflow they already use.
  • For high-volume form processing, the claimed 64% per-form cost reduction would compound into substantial annual savings even with a human reviewer in the loop.
  • Human verification remains essential: even the best model left 11 fields blank and put errors in 2, so the pipeline is semi-automated, not autonomous.
  • The architecture is model-agnostic, so an organization can swap the LLM and use the 29-field benchmark to trade model cost against field-level accuracy.
  • Because the completed form returns through the same email channel, the email thread becomes an audit trail of request, context, and output.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper, the same email-as-prompt pattern should transfer to other structured documents such as invoices, applications, and claims, as long as the target fields are well defined and the context documents are supplied.
  • Beyond the paper, the cost model depends on a fixed five-minute review, so the natural next measurement is actual review time across staff; the break-even point moves directly with that number.
  • A testable extension would be a low-confidence rule: when the model is unsure of a field, leave it blank and flag it for the reviewer, pushing real accuracy toward the optimal-result baseline.
  • Beyond the paper, sending identity documents and forms through a cloud email service and OCR provider moves sensitive administrative data outside the organization, a privacy consideration the paper mentions only briefly.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 7 minor

Summary. The paper proposes an architecture that lets non-technical administrative staff delegate form-filling tasks to an LLM-based system through ordinary email, with attachments processed by OCR and the email body used as a natural-language prompt. A proof of concept is implemented and tested on a 29-field Spanish administrative form with seven LLMs. The authors report that the best model (llama-4-maverick) fills 16 fields correctly, that OCR processing takes 7.8 seconds, and that with 5 minutes of human supervision the workflow reduces staff time by a factor of three to four and cuts cost by 64.6% relative to manual completion. The paper concludes that email-based LLM integration is a viable and cost-effective way to democratize administrative automation.

Significance. If the quantitative claims were supported by rigorous measurement, the contribution would be useful: the email-as-interface idea is intuitive, the architectural description is clear, and the open-source proof of concept is a strength that allows others to reproduce and extend the work. The paper also makes an honest effort to compare several LLMs and to distinguish intentional blank fields from errors. However, the central claims of viability and cost-effectiveness currently rest on a single un-replicated trial on one form and on an unmeasured 5-minute supervision assumption. As a result, the significance is real but provisional; the paper would need a substantially stronger evaluation before the headline conclusions can be accepted.

major comments (4)
  1. [Section V-C, Table II] The 5-minute human supervision figure is asserted rather than measured, and it is the load-bearing assumption for both the claimed 3-4x staff-time reduction and the 64.6% cost saving. The paper's own Table I shows that the best model leaves 11 fields blank and fills 2 incorrectly, so a reviewer must check 29 fields against source documents and decide whether blanks are intentional or missing. No protocol, timing data, or user study supports the claim that this can be done reliably in 5 minutes. The sensitivity is substantial: if supervision takes 10 minutes, the cost saving drops to roughly 29% and the time reduction falls below 2x. The wage assumption is also inconsistent with the cited source: the text uses approximately EUR 40,000, but reference [30] (INE Wage Structure Survey) reports an average annual salary of EUR 28,049.94, which would lower the manual baseline and further reduce the savings. The authors should either measure supervision time, present a sensitivity analysis over supervision time and wage, or substantially soften the economic claims.
  2. [Section V-A, Table I] The evaluation is a single-trial measurement on one form, with no repeated runs, no confidence intervals, and no comparison against a baseline such as manual completion accuracy or a simpler rule-based automation. The '7.8 +/- 0.3 seconds' figure is reported as the time for AWS Rekognition to process the document, not the end-to-end time of the full pipeline including LLM inference and form population, yet the abstract states that the system 'can complete complex administrative forms in under 8 seconds.' The paper should report end-to-end latency separately from OCR time, and should provide repeated measurements and uncertainty estimates for the field-level accuracy results.
  3. [Abstract, Table I] The abstract claims the system 'accurately filled 16 out of 29 form fields' and that this 'establish[es] email-based LLM integration as a viable and cost-effective approach.' However, Table I shows that the best model also produced 2 incorrect fields and 11 blanks, and the 'Optimal result' row itself leaves 12 fields blank because not all fields are answerable from the provided documents. Reporting only the 16 correct fields overstates practical completeness: for a 29-field form with 2 errors and 11 blanks, the system still requires substantial human correction and verification. The viability claim should be qualified to reflect that the current system is semi-automated and requires careful human review, rather than presenting 16/29 as near-complete automation.
  4. [Section V-C] The estimated cloud processing cost of EUR 0.10 per form is said to be 'based on our simulations,' but no details are given about the OCR and LLM pricing assumptions, the number of documents processed, or the distribution of tokens. Since this cost is a component of the claimed 64.6% saving, the figure needs to be justified with a concrete pricing model or a range of estimates.
minor comments (7)
  1. [Section I] The introduction refers to 'Optical Content Recognition' instead of 'Optical Character Recognition.'
  2. [Abstract] The sentence '64% relatively reduced the total cost per processed form to manual completion' is ungrammatical and should be rewritten, for example as 'reduced the total cost per processed form by 64% relative to manual completion.'
  3. [Table II] The column header 'Cost [ C]' appears to be a character-encoding artifact; it should read 'Cost [EUR]' or 'Cost [euros].' Similar encoding artifacts appear elsewhere in the text.
  4. [Section V-A] The text says llama-4-maverick achieved '16 correct fields (94% of optimal benchmark),' but the optimal row in Table I has 17 correct fields, so the correct percentage is 16/17 = 94.1%; please clarify the calculation.
  5. [Section V-A] The description of performance tiers groups llama-3.3-70b-instruct with qwen-qwq-32b as 'lower accuracy outcomes (10 correct),' but llama-3.3 also has 8 incorrect fields, which makes it qualitatively worse than qwen-qwq's 10 incorrect fields; the tier description should reflect the error counts as well.
  6. [Reference [30]] The text in Section V-C cites reference [30] as supporting an annual salary of approximately EUR 40,000, but the cited INE survey reports an average annual salary of EUR 28,049.94. Either cite a different source for the EUR 40,000 figure or revise the text to match the cited data.
  7. [Section VII] The claim that 'when this paper was written, MCPs were not yet available or mature enough to use in practice' is questionable for a 2025 manuscript, since Model Context Protocol had already been publicly introduced by then; please clarify the specific version or maturity threshold intended.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the technical evaluation is empirical and self-contained, and the economic figures are one-way arithmetic from stated assumptions rather than predictions that feed back into those assumptions.

full rationale

The paper's central technical claim—that an email-plus-OCR-plus-LLM pipeline can populate administrative forms—rests on direct measurements reported in Table I (correct/incorrect/blank fields per model) and the measured 7.8s AWS Rekognition processing time, with no fitted parameter reused as a predicted quantity. The architecture is described independently of the results. The Section V-C time and cost savings are deterministic arithmetic consequences of explicitly stated inputs (15–20 min manual completion, ~5 min supervision, ~€40,000 annual salary, €0.10 cloud cost), so the 3–4x staff-time factor and 64.6% cost reduction are not independently measured and are sensitive to those assumptions; this is a validity and robustness limitation, not a circularity, because no equation is definitionally identical to the claimed conclusion and the assumptions are not themselves derived from the savings figures. There are no self-citations carrying the argument, no imported uniqueness theorem, no ansatz smuggled in via citation, and no renamed known result. The paper is therefore self-contained against external benchmarks for its empirical core, and no step in the derivation chain reduces to its own inputs by construction.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central economic claims are computed from assumed time and wage inputs rather than measured outcomes. The system architecture itself introduces no new theoretical entities, but it relies on domain assumptions about LLM, OCR, and form representativeness that are not independently validated in the paper.

free parameters (4)
  • human_supervision_time_per_form = 5 minutes (assumed)
    Section V-C uses 5 minutes of human review per document to compute the 3-4x time reduction and the 1.70 euro cost per form. This is an estimate, not a measured duration.
  • manual_completion_time_per_form = 15 minutes (text states 15-20)
    Table II uses 15 minutes for the manual baseline, giving 4.81 euro at 19.23 euro/hour. The text says 15-20 minutes, so the cost and savings figures depend on the lower bound.
  • effective_hourly_wage = 19.23 euros/hour
    Derived from an assumed 40,000 euros annual gross plus social security (Section V-C). The cited INE source reports an average of 28,049.94 euros, so this input is not directly supported.
  • cloud_processing_cost_per_form = 0.10 euros (approximate)
    Section V-C: 'maximum estimated cloud processing cost (0.10 euros per form, based on our simulations)'. No cost breakdown is provided.
assumptions (5)
  • domain assumption A large language model can map OCR-extracted text to the correct fields of an unseen administrative form when given a textual instruction and the attachment content.
    Invoked in Section III-C (LLM responsibilities) and Section IV (proof of concept); no controlled test isolates the LLM's contribution from the Python filler or the OCR text quality.
  • domain assumption AWS Rekognition's OCR output is faithful enough for the downstream LLM and form-filler.
    Section III-B; no OCR accuracy measurement or comparison to Textract/Tesseract/EasyOCR is reported.
  • domain assumption The single 29-field Spanish form used in the evaluation is representative of common administrative forms and therefore supports general claims about viability.
    Section IV and Table I; only one form was tested, with no multi-form or multi-language experiment.
  • domain assumption Five minutes of human supervision per document is a realistic estimate of the review time required after automated filling.
    Section V-C, paragraph 2; this number is stated without measurement, yet it drives the 3-4x time reduction and the 64% cost saving.
  • domain assumption The cost model inputs (wage, manual completion time, cloud cost) are adequate to support the economic comparison.
    Section V-C; wage is derived from an assumed 40,000 euros annual cost while the cited INE source reports 28,049.94 average salary, and cloud cost is given as 'approximately 0.10 euros per form based on simulations'.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Email as the Interface to Generative AI Models: Seamless Administrative Automation." pith.science (2026). https://pith.science/paper/IE5QTWKP

@misc{pith2026250623850,
  author       = {Pith},
  title        = {Pith review of: Email as the Interface to Generative AI Models: Seamless Administrative Automation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IE5QTWKP}},
  note         = {Machine review of arXiv:2506.23850}
}
read the original abstract

This paper introduces a novel architectural framework that integrates Large Language Models (LLMs) with email interfaces to automate administrative tasks, specifically targeting accessibility barriers in enterprise environments. The system connects email communication channels with Optical Character Recognition (OCR) and intelligent automation, enabling non-technical administrative staff to delegate complex form-filling and document processing tasks using familiar email interfaces. By treating the email body as a natural language prompt and attachments as contextual information, the workflow bridges the gap between advanced AI capabilities and practical usability. Empirical evaluation shows that the system can complete complex administrative forms in under 8 seconds of automated processing, with human supervision reducing total staff time by a factor of three to four compared to manual workflows. The top-performing LLM accurately filled 16 out of 29 form fields and reduced the total cost per processed form by 64% relative to manual completion. These findings demonstrate that email-based LLM integration is a viable and cost-effective approach for democratizing advanced automation in organizational settings, supporting widespread adoption without requiring specialized technical knowledge or major workflow changes. This aligns with broader trends in leveraging LLMs to enhance accessibility and automate complex tasks for non-technical users, making technology more inclusive and efficient.

Figures

Figures reproduced from arXiv: 2506.23850 by the authors.

Figure 1
Figure 1. Example email workflow for administrative automation: [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. System workflow architecture showing the integration [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. End-to-end workflow architecture for autonomous [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Example test performed with the proposed workflow [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 27 canonical work pages

  1. [30]

    DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

    DeepSeek-AI et al. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning. 2025. arXiv: 2501.12948 [cs.CL]. URL: https://arxiv.org/abs/2501.12948

    arXiv 2025

  2. [1]

    LLM-Based Framework for Administrative Task Automation in Healthcare

    Senay A. Gebreab et al. “LLM-Based Framework for Administrative Task Automation in Healthcare”. In: 2024 12th International Symposium on Digital Forensics and Security (ISDFS). 2024, pp. 1–7. DOI: 10.1109/ISDFS60797.2024.10527275

    arXiv 2024

  3. [2]

    AutoDroid: LLM-powered Task Automation in Android

    Hao Wen et al. “AutoDroid: LLM-powered Task Automation in Android”. In: Proceedings of the 30th Annual International Conference on Mo- bile Computing and Networking . ACM MobiCom ’24. Washington D.C., DC, USA: Association for Computing Machinery, 2024, pp. 543–557. ISBN : 9798400704895. DOI: 10.1145/3636534.3649379. URL: https://doi.org/10.1145/3636534.3649379

    arXiv 2024

  4. [3]

    Utilizing Large Language Models for Advanced Service Manage- ment: Potential Applications and Operational Chal- lenges

    Sudhakar Reddy Peddinti et al. “Utilizing Large Language Models for Advanced Service Manage- ment: Potential Applications and Operational Chal- lenges”. In: Journal of Science & Technology 4.2 (Mar. 2023). Available at SSRN: https://ssrn.com/ abstract = 5119925 or http : / / dx . doi . org / 10 . 2139 / ssrn . 5119925. DOI: 10 . 2139 / ssrn . 5119925.URL:...

    work page 2023

  5. [4]

    Potentials and Challenges of Large Language Models (LLMs) in the Context of Administrative Decision-Making

    Paulina Jo Pesch. “Potentials and Challenges of Large Language Models (LLMs) in the Context of Administrative Decision-Making”. In: European Journal of Risk Regulation 16.1 (2025), pp. 76–95. DOI: 10.1017/err.2024.99

    doi:10.1017/err.2024.99 2025

  6. [5]

    The digital workplace is key to digital innovation

    Kristine Dery, Ina M. Sebastian, and Nick van der Meulen. “The digital workplace is key to digital innovation”. English. In: MIS Quarterly Executive 16.2 (2017), pp. 135–152. ISSN : 1540-1960. 7

    work page 2017

  7. [6]

    Ar- tificial Intelligence Scribe and Large Language Model Technology in Healthcare Documentation: Advantages, Limitations, and Recommendations

    SA Ess, AJ Mackey, and DE Yarowsky. “Ar- tificial Intelligence Scribe and Large Language Model Technology in Healthcare Documentation: Advantages, Limitations, and Recommendations”. In: Plastic and Reconstructive Surgery Global Open 13.1 (Jan. 2025), e6450. DOI: 10 . 1097 / GOX . 0000000000006450

    work page 2025

  8. [7]

    An Empirical Study on Challenges for LLM Application Developers

    Xiang Chen et al. An Empirical Study on Challenges for LLM Application Developers. 2025. arXiv: 2408. 05002 [cs.SE]. URL: https://arxiv.org/abs/2408. 05002

    work page 2025

Show all 33 references
  1. [8]

    Ruoyan Kong, Haiyi Zhu, and Joseph A. Konstan. Organizational Bulk Email Systems: Their Role and Performance in Remote Work . 2023. arXiv: 2308. 05085 [cs.HC]. URL: https://arxiv.org/abs/2308. 05085

    2023

  2. [9]

    Craw- ford

    Nathaniel Josephs, Sida Peng, and Forrest W. Craw- ford. Communication network dynamics in a large organizational hierarchy. 2024. arXiv: 2208.01208 [stat.AP]. URL: https : / / arxiv. org / abs / 2208 . 01208

    2024 arXiv

  3. [10]

    E-Mail Assistant – Automation of E-Mail Handling and Management using Robotic Process Automation

    Arpit Khare et al. “E-Mail Assistant – Automation of E-Mail Handling and Management using Robotic Process Automation”. In: 2022 International Con- ference on Decision Aid Sciences and Applications (DASA). IEEE, Mar. 2022, pp. 511–516. DOI: 10. 1109/dasa54658.2022.9765017. URL:...

    2022

  4. [11]

    Understanding Emails and Drafting Re- sponses – An Approach Using GPT-3

    Jonas Thiergart, Stefan Huber, and Thomas ¨Ubellacker. Understanding Emails and Drafting Re- sponses – An Approach Using GPT-3 . 2021. arXiv: 2102.03062 [cs.AI]. URL: https://arxiv.org/abs/ 2102.03062

    2021 arXiv

  5. [12]

    Smart Reply: Automated Re- sponse Suggestion for Email

    Anjuli Kannan et al. Smart Reply: Automated Re- sponse Suggestion for Email . 2016. arXiv: 1606 . 04870 [cs.CL]. URL: https://arxiv.org/abs/1606. 04870

    2016

  6. [13]

    Field Label Prediction for Autofill in Web Browsers

    Joy Bose. Field Label Prediction for Autofill in Web Browsers. 2019. arXiv: 1912.08809 [cs.HC]. URL: https://arxiv.org/abs/1912.08809

    2019 arXiv

  7. [14]

    Aveni, Armando Fox, and Bj ¨orn Hart- mann

    Timothy J. Aveni, Armando Fox, and Bj ¨orn Hart- mann. OmniFill: Domain-Agnostic Form Filling Suggestions Using Multi-Faceted Context . 2023. arXiv: 2310.17826 [cs.HC]. URL: https://arxiv. org/abs/2310.17826

    2023 arXiv

  8. [15]

    Amazon Textract: Automat- ically Extract Printed Text, Handwriting, Layout Elements, and Data from Any Document

    Amazon Web Services. Amazon Textract: Automat- ically Extract Printed Text, Handwriting, Layout Elements, and Data from Any Document . https : / / aws.amazon.com/textract/. Accessed: 2025-05-29. 2018

    2025

  9. [16]

    Amazon Rekognition: Auto- mate and Lower the Cost of Your Image Recognition and Video Analysis with ML

    Amazon Web Services. Amazon Rekognition: Auto- mate and Lower the Cost of Your Image Recognition and Video Analysis with ML . https://aws.amazon. com/rekognition/. Accessed: 2025-05-29. 2016

    2025

  10. [17]

    EasyOCR: Ready-to-use OCR with 80+ Languages Supported

    JaidedAI. EasyOCR: Ready-to-use OCR with 80+ Languages Supported. https://github.com/JaidedAI/ EasyOCR. Accessed: 2025-05-29. 2020

    2025

  11. [18]

    Tesseract Open Source OCR Engine

    Ray Smith, Zdenko Podobny, and Tesseract OCR Contributors. Tesseract Open Source OCR Engine . https://github.com/tesseract- ocr/tesseract. Version 5.5.1. Google Inc., 2025. URL: https://github.com/ tesseract-ocr/tesseract

    2025

  12. [19]

    Amazon WorkMail

    Amazon Web Services, Inc. Amazon WorkMail . https : / / aws . amazon . com / workmail/. Accessed: 2025-05-18. 2025

    2025

  13. [20]

    Agent-S: LLM Agentic workflow to automate Standard Operating Procedures

    Mandar Kulkarni. Agent-S: LLM Agentic workflow to automate Standard Operating Procedures . 2025. arXiv: 2503.15520 [cs.HC]. URL: https://arxiv. org/abs/2503.15520

    2025 arXiv

  14. [21]

    TA.0521.1 Simplificada (V9)

    Ministerio de Justicia. TA.0521.1 Simplificada (V9). Formulario oficial de la Seguridad Social de Espa ˜na

  15. [22]

    Formulario de solicitud de consulta de documentos del Archivo General

    Ministerio de Justicia. Formulario de solicitud de consulta de documentos del Archivo General . For- mulario oficial del Ministerio de Justicia de Espa ˜na

  16. [23]

    Llama 4 Maverick 17B-128E Instruct

    Meta AI. Llama 4 Maverick 17B-128E Instruct . Model release date: April 5, 2025. Llama 4 Mav- erick is a 17B parameter, 128-expert, natively mul- timodal large language model released under the Llama 4 Community License. Knowledge cutoff: August 2024. 2025. URL: https://huggin...

    2025

  17. [24]

    mjusticia

    URL: https : / / sede . mjusticia . gob . es / es / TramitesSede / Documents / 1292429619664 - Formulario de solicitud de consulta . pdf (visited on 06/08/2025)

    2025

  18. [25]

    Gemini 2.5: Our most intelligent AI model

    Google DeepMind. Gemini 2.5: Our most intelligent AI model. Last updated March 26. Introduces Gem- ini 2.5, a state-of-the-art ”thinking model” designed for advanced reasoning and complex problem- solving. Mar. 2025. URL: https : / / blog . google / technology / google - deepm...

    2025

  19. [26]

    Llama 4 Scout 17B-16E Instruct

    Meta AI. Llama 4 Scout 17B-16E Instruct . Model release date: April 5, 2025. Llama 4 Scout is a 17B parameter, 16-expert, natively multimodal large language model released under the Llama 4 Com- munity License. Knowledge cutoff: August 2024

    2025

  20. [27]

    co / meta - llama / Llama-4-Scout-17B-16E-Instruct

    URL: https : / / huggingface . co / meta - llama / Llama-4-Scout-17B-16E-Instruct

  21. [28]

    Llama 3.3 70B Instruct

    Meta AI. Llama 3.3 70B Instruct . Model release date: December 6, 2024. Llama 3.3 is a 70B param- 8 eter, instruction-tuned, multilingual large language model released under the Llama 3.3 Community License. Knowledge cutoff: December 2023. 2024. URL: https://huggingface.co/met...

    2024

  22. [29]

    Introducing GPT-4.1 in the API

    OpenAI. Introducing GPT-4.1 in the API . An- nounces the release of GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano, with major improvements in coding, instruction following, and long context handling. June 2025. URL: https://openai.com/index/gpt-4-1/

    2025

  23. [32]

    QwQ-32B: Embracing the Power of Reinforcement Learning

    Qwen Team. QwQ-32B: Embracing the Power of Reinforcement Learning . Mar. 2025. URL: https:// huggingface.co/Qwen/QwQ-32B

    2025

  24. [33]

    Wage Struc- ture Survey (Encuesta de Estructura Salarial)

    Instituto Nacional de Estad ´ıstica (INE). Wage Struc- ture Survey (Encuesta de Estructura Salarial) . Av- erage annual salary in 2023: 28,049.94 euros per worker. Published: 28/05/2025. Instituto Nacional de Estad´ıstica (INE). 2023. URL: https://www.ine.es/ dyngs/INEbase/en/...

    2023

  25. [2025]

    seg - social

    URL: https : / / sede . seg - social . gob . es / wps / wcm / connect / sede / 9b1bd995 - f5f7 - 4dba - aba1-3a8e04105d79/TA 0521 1%2BSimplificada% 2B%28V9%29.pdf?MOD=AJPERES (visited on 06/08/2025)

    2025

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.