# Apodex 1.1 Advances Agentic Models with Async Teams and Open Mini

> The release combines asynchronous Agent Team coordination, a locally deployable 35B model, and an open harness to enable executable long-horizon work in finance, research, and professional domains.

*Published 2026-08-25 · By Marcus Vance*

Apodex 1.1 is a family of models that scales agentic intelligence for complex professional, scientific research, and financial tasks with asynchronous Agent Team capabilities.

Apodex announced the Apodex 1.1 family on August 24, 2026, through its technical report and official X channels. The models target complex tasks that require sustained coordination and verifiable execution rather than one-shot outputs.

## Background on Agentic Intelligence Scaling

Earlier agentic systems frequently delivered static summaries that lacked direct executability. Apodex 1.1 focuses on grounding performance in actual deliverables through structured environments and coordinated agent behaviors.

The approach rests on two documented scaling paths. Environment Scaling supplies executable file, search, and code environments. Agentic Coordination Scaling covers task decomposition, parallel sub-agents, and asynchronous integration of results.

## New Features in the Apodex 1.1 Family

The primary Apodex 1.1 model is offered through an online workbench and API. The 35B-parameter Apodex 1.1 Mini version is released as open weights for local deployment on standard hardware.

FrontierAgent was released concurrently on GitHub as an open-source runtime and command-line TUI. It supports both ReAct and Agent Team modes for long-horizon research and file-based tasks.

## Technical Details of the Scaling Paths

Environment Scaling provides agents with direct access to executable contexts. Agents can run code, query external data, and manipulate files within controlled settings. This setup supports scientific research workflows that require iterative testing and data organization.

Agentic Coordination Scaling decomposes complex objectives into sub-tasks. Multiple sub-agents operate in parallel on assigned portions. Asynchronous mechanisms collect and integrate outputs over extended periods without requiring constant oversight.

## Benchmark Results and Performance Metrics

Evaluations on domain-specific benchmarks show measurable gains when the Agent Team configuration is applied.

Benchmark scores achieved by Apodex 1.1 using the Agent Team configuration, sourced from official announcements and reports.BenchmarkApodex 1.1 Agent Team ScoreFrontierFinance54.3FrontierScience-Research63.3APEX-Agents38.5GDPVal78.8

The Agent Team configuration delivers an improvement of 4.1 to 9.3 points over the ReAct baseline across the evaluated benchmarks.

## Open Releases and Accessibility

The Apodex 1.1 Mini weights are available on Hugging Face. This distribution supports local inference and integration into environments where data privacy or offline operation is required.

FrontierAgent is distributed via GitHub and functions as a terminal-based workbench with built-in evaluation capabilities. Users can run the harness locally to test both ReAct and Agent Team workflows.

## Implications for Markets and Stakeholders

Finance professionals can apply the reported 78.8 GDPVal win rate to support analysis tasks. Research teams may utilize the FrontierScience-Research score of 63.3 to gauge suitability for extended scientific workflows.

The dual availability of cloud API access and local open weights gives organizations flexibility based on infrastructure and compliance needs. Open components of the release may accelerate experimentation by external developers.

## Official Statements and Reactions

The Apodex official account posted statements on X describing the model family and its intended applications.

> We’re excited to introduce Apodex 1.1, our new model family built to scale agentic intelligence for professional work.Apodex, Official account

A follow-up statement noted frontier-level agentic performance across professional work, scientific research, financial analysis, and deep search.

## What's Next for Apodex and Agentic Systems

Subsequent iterations may extend the range of supported environments and refine coordination protocols for tasks spanning even longer time horizons.

The open-source FrontierAgent framework may receive community updates that add new workflow options or improved evaluation tools.

## Agent Team Mode Workflow

- Decompose the overall task into manageable sub-tasks based on the requirements.
- Deploy parallel sub-agents to handle individual sub-tasks simultaneously.
- Allow asynchronous execution and result collection over the task duration.
- Integrate outputs from sub-agents to form a cohesive final deliverable.
- Verify the results through executable checks and validation steps.

## Sources

1. [Apodex 1.1 is a general-purpose model and execution system that scales agentic intelligence for complex work. The 35B-parameter Apodex 1.1 Mini further retains strong working capability in a locally deployable form. It reports 54.3 on FrontierFinance and 63.3 on FrontierScience-Research with Agent Team.](https://framerusercontent.com/assets/vnib7j93v0EP1kkb4GmFU8WEU.pdf)
2. [Apodex 1.1 with Agent Team achieves 38.5 on APEX-Agents and 78.8 win rate on GDPVal. Agent Team adds 4.1–9.3 points over ReAct.](https://x.com/Apodex_AI/status/2091916794319819104)
3. [Official announcement of Apodex 1.1 release.](https://x.com/Apodex_AI/status/2091916791308313018)
4. [FrontierAgent is an open-source agent runtime, terminal product, and evaluation suite for long-horizon research and file-based work. The `frontier-agent` TUI ships two native workflows: ReAct and Agent Team.](https://github.com/ApodexAI/FrontierAgent)
5. [Apodex released Apodex 1.1, a new family of models scaling agentic intelligence for complex professional, scientific research, and financial tasks with asynchronous Agent Team capabilities. Includes open-weight 35B…](https://apodex.ai)

---
Source: https://aiintelreport.com/frontier-models/apodex-1-1-agentic-model-family-release
Index: https://aiintelreport.com/llms.txt · Full text: https://aiintelreport.com/llms-full.txt