Apodex 1.1: From Release to Hands-On Tests

Episode 1 · Apodex Launches 1.1 with Open-Source FrontierAgent Framework and 35B Mini Model (2026-09-08, 14 posts)

Apodex has released version 1.1, open-sourcing the FrontierAgent execution framework alongside a 35B-parameter 1.1 mini model. The entire release rests on one thesis: the meaningful unit for measuring AI capability is end-to-end completion of a real job, producing checkable artifacts like documents and code, not merely answering questions correctly—"a correct answer can still be a failed task." That thesis, paired with a verifiable-deliverable mechanism, makes this a release worth watching for anyone focused on trustworthy agent output.

Confirmed

• FrontierAgent is open-sourced on GitHub; it launches with a single command on macOS/Linux, with no pre-installation or hard Docker dependency; it features a native command-line TUI with single-agent and Agent Team modes, and currently has around 2.4k stars
• Apodex 1.1 mini is a 35B-parameter open-weights model on Hugging Face, with FP8, NVFP4, GPTQ-Int4 and other quantized variants, and can be deployed locally alongside FrontierAgent
• The full model powers the online web workspace, with a free two-week API trial for new users
• Official figures place 1.1 with Agent Team mode in the current frontier tier of agent systems on benchmarks including APEX-Agents, GDPval, FrontierScience, and Humanity's Last Exam; per reposts, the Apodex 1.1 paper (arXiv:2608.23283) and Agent Team evaluation results have been added to Papers with Code, with the Agent team achieving a 63.3% pass rate on FrontierScience
• By design, "generating results" and "checking results" are separate steps (Statement Review): key conclusions are double-checked by an independent stage before delivery, flagging weak evidence, citation mismatches, and calculation conflicts while retaining an audit trail
• Deep Discover mode has a built-in asynchronous Agent Team: the model itself decides task decomposition, the number of sub-agents, and when to aggregate; sub-agents run in parallel and stream results back to a shared task state; new files or changed requirements can be injected mid-run, with the system preserving valid work and re-planning only affected parts—the official demo is a global EV battery supplier risk assessment
• Official demos handle real professional files directly, e.g. reading raw clinical data tables for cleaning and survival analysis with Kaplan-Meier curves, plus powder XRD indexing and batch protein docking

Why it matters

• The release shifts from "telling you how" to "working alongside you," making verifiable deliverables—not Q&A accuracy—the capability standard, targeting the trustworthiness of agent output in real professional settings
• The Statement Review mechanism separating generation from verification was flagged by reviewer aakashgupta as the most underrated capability, and a key safeguard for using agent output in actual work
• The combination of an open-source framework and open 35B weights lowers the barrier to local reproduction, with the mini model able to run full workflows locally alongside FrontierAgent

• Apodex 1.1 Agent Team hits 63.3% pass rate on FrontierScience-Research, paper lands on Papers with Code — NielsRogge · 2026-09-08
• Apodex 1.1 Argues a Right Answer Can Still Be a Failed Task — aakashgupta · 2026-09-09
• Apodex 1.1 argues a right answer can still be a failed task — aakashgupta · 2026-09-09
• Inside Apodex: an asynchronous Agent Team running on a live task board — aakashgupta · 2026-09-09
• Apodex agents absorb mid-task requirement changes with a partial replan — aakashgupta · 2026-09-09
• Apodex 1.1 review: Statement Review independently verifies agent conclusions — aakashgupta · 2026-09-09
• Apodex 1.1 Demos Hit Real Clinical Data, and Statement Review May Be the Sleeper Feature — aakashgupta · 2026-09-09
• Apodex Claims 1.1 With Agent Team Sits in the Frontier Tier; 35B Mini Runs Locally — aakashgupta · 2026-09-09
• Apodex 1.1 Mini Weights Hit Hugging Face; Full Model API Free for Two Weeks — aakashgupta · 2026-09-09
• Apodex Open-Sources FrontierAgent: One-Command Local Agent Framework With Agent Team Mode — aakashgupta · 2026-09-09
• FrontierAgent splits generation from verification, flagging weak evidence before delivery — mhdfaran · 2026-09-09
• Apodex open-sources FrontierAgent: local multi-agent TUI framework with ReAct and Agent Team modes — mhdfaran · 2026-09-09
• Apodex open-sources FrontierAgent agent framework and ships 35B open-weight Apodex 1.1 mini — mhdfaran · 2026-09-09
• Apodex 1.1 launches with open-weight 35B mini model for local deployment — mhdfaran · 2026-09-09

Episode 2 · Hands-On: Apodex 1.1 Rearranges Workflows When Requirements Change Mid-Task (2026-09-09, 3 posts)

Users testing Apodex 1.1 found it can automatically reorganize research pipelines and rerun multi-agent task boards when requirements are changed mid-task, demonstrated on a complex open-source model comparison benchmark.

• User Puts Apodex 1.1 to the Test: It Reorganizes Work Mid-Run When Requirements Change — mhdfaran · 2026-09-09
• Hands-on with Apodex 1.1: changing requirements mid-run reorganizes the multi-agent task board — mhdfaran · 2026-09-09
• Apodex 1.1 rebuilds deep research around files, method selection and verifiable conclusions — mhdfaran · 2026-09-09