Komori
Abeja (5574) · InitiativeAdvanced

Expanding LLM services on the ABEJA Platform and enterprise deployment

First written in FY2023-Q3. 17 of 19 entries say "Advanced", first in FY2024-Q2. Latest entry: 2026-10-15.

19entries
4years
8from disclosures
Share on X
How it moved
ProgressHeldSoftenedWorseFirst statedFY2023-Q3 – FY2023AdvancedFY2024-Q2 – 2026-10-152023: 220232024: 320242025: 220252026: 122026
Same theme, other companies
Development and provision of advanced technology solutions, including generative AIA theme 52 companies write about · 0 as a headwind · 62 as a tailwind

40 other companies' latest entry also says "Advanced".

Ridges = years it wrote progress, valleys = years it wrote worse. Most hopeful companies at the back. Drag to turn, scroll to zoom, right-drag to pan. On a phone, two fingers turn and pinch to zoom; one finger scrolls the page.
Showing 14 of the 52 companies compared. The sentences above are computed from all 52.
Full recordFilings, disclosures and news, newest first
  1. FY2026-H1AdvancedAnnual report

    「During the interim accounting period, our main initiative was the NEDO project under the Ministry of Economy, Trade and Industry's GENIAC framework, titled "Research and Development on Foundation Models with Long-Context Capabilities and the Construction of AI Agents" (selected in July 2025 and completed in February 2026).」

    Specific progress was reported in LLM and AI-agent activities: the NEDO project on foundation models with long-context capabilities and the construction of AI agents was completed in February 2026.
  2. FY2025AdvancedAnnual report

    「During the fiscal year, we pursued enterprise projects and public-sector projects in parallel and worked to accelerate social implementation. In research and development, we built a high-accuracy compact LLM under the NEDO framework and confirmed a breakthrough in the cost-to-accuracy trade-off.」

    For LLMs, the company pursued enterprise projects and public-sector projects in parallel and reported building a high-accuracy compact LLM and confirming an improvement in the cost-to-accuracy trade-off.
  3. FY2025-H1AdvancedAnnual report

    「We continued to promote LLM-related activities as a focus area during the interim accounting period. Under NEDO's "Development of Competitive Generative AI Foundation Models (Subsidy)" project, selected in October 2024, we are conducting research and development on "high-performance, compact models with reduced parameter counts," drawing on knowledge gained from the previously selected project. The project is progressing smoothly, with results including the development of a 32B compact model that has outperformed GPT-4 on multiple general-purpose language-performance benchmarks. We also focused on developing a Japanese-language medical LLM, in which we are participating as a joint research and development institution. In addition, we are making steady progress toward the social implementation of LLMs by creating LLM use cases with enterprise companies and implementing the compact model described above in edge environments to improve usability.」

    Specific progress was reported in LLM-related activities, including compact-model research and development under the NEDO project selected in October 2024, development of a 32B model that outperformed GPT-4 on multiple benchmarks, a Japanese-language medical LLM, enterprise use cases and edge implementation.
  4. FY2024AdvancedAnnual report

    「We continued to focus on LLM-related activities. As an NEDO project, we conducted research and development on "Japanese-language LLMs and related technologies (RAG and agents)" from February through August 2024 following its selection (the "NEDO Phase 1 Project"). From the next fiscal year onward, LLM-related activities will remain a focus area, and we will continue to promote research and development in addition to providing services to customer companies.」

    The NEDO Phase 1 Project was completed following its selection in February 2024 and implementation through August. The company indicated that it will continue to focus on LLMs, customer services and research and development from the next fiscal year onward.
  5. FY2024-Q3AdvancedAnnual report

    「Since its selection in February 2024, we have been conducting research and development on Japanese-language LLMs and related technologies, including RAG and agents, as part of an NEDO project. We are currently working to improve the accuracy of the foundation model and have also undergone GENIAC (Generative AI Accelerator Challenge) performance evaluations, with progress proceeding smoothly. As the project is scheduled to reach a milestone in August this year, we plan to release the models and other outputs after its completion.」

    Specific progress and plans were reported for research and development on Japanese-language LLMs and related technologies following NEDO's selection, including improvements in foundation-model accuracy, GENIAC performance evaluations, the August 2024 project milestone and plans to release the models after completion.
  6. FY2024-Q2AdvancedAnnual report

    「We also applied for the New Energy and Industrial Technology Development Organization (NEDO)'s project, "Post-5G Information and Communications Systems Infrastructure Enhancement Research and Development Project / Development of Post-5G Information and Communications Systems." Effective 2 February 2024, our proposal, "A General-Purpose LLM as the Basis for Specialized Models for the Social Implementation of LLMs," was selected.」

    NEDO selected the research and development proposal for a general-purpose LLM aimed at the social implementation of LLMs, marking progress from the previous LLM deployment policy to a specific research and development phase.
  7. FY2023RestatedAnnual report

    「We are also proceeding with the deployment of our proprietary large language model, the ABEJA LLM Series, on ABEJA Platform.」

    The policy of deploying LLMs on Platform remains in place. The annual report also identifies large language models as a primary research and development focus, but does not state that deployment has been completed.
  8. FY2023-Q3First statedAnnual report

    「During the third-quarter accounting period, we expanded the large language model services on ABEJA Platform. Our LLMs primarily target enterprises, and we are advancing their deployment toward implementation and operation.」

    The expansion of LLM services and their deployment toward implementation and operation for enterprise customers was recorded for the first time.
Moved together

"The Five Layers of ABEJA Platform and Investment in Feature Development for LLMs" moved the same way as this thread, in the same document, in 9 of its 19 moves.

987654★123456
  1. ★This thread
  2. 1The Five Layers of ABEJA Platform and Investment in Feature Development for LLMs9×
  3. 2Steady business expansion8×
  4. 3Increase in Stated Capital and Capital Surplus through the Exercise of Stock Acquisition Rights7×
  5. 4Expanding the scope of LLM applications in AI robotics6×
  6. 5Return to Operating Profitability and Profit Generation5×
  7. 6Human–AI Collaboration through Human-in-the-Loop on the ABEJA Platform4×

★ is this thread; numbers match the list above. The number on a line is how many documents both moved the same way in; lines between the other threads count the documents they moved together in too. Threads that move together often sit closer.

When this could move next
Abeja's next results are scheduled for 2026-10-15. Read the earnings preview It last moved in the Financial Results Presentation of 2026-10-15: "Advanced".
Expanding LLM services on the ABEJA Platform and enterprise deployment | ABEJA,Inc. (5574) | Komori