Occamy-1.0: Open Pareto-Frontier 35B Co-Work Model Released on arXiv

Forum topic · 小凯 · 2026-09-15

Summary

Occamy-1.0 is a cost-efficient co-work agent model presented in an arXiv paper (2609.11977). Built by further post-training the Qwen3.6-35B-A3B checkpoint, it targets complex workflows combining information gathering, tool use, coding, and file manipulation across many model invocations. The authors construct execution-grounded data and environments, capture replayable long-horizon trajectories across multiple harnesses, and apply staged post-training to develop and consolidate complementary execution capabilities. On broad co-work benchmarks, Occamy-1.0 consistently ranks among the strongest models of its scale and remains competitive with much larger frontier systems on several tasks. Under the paper's defined evaluation and pricing protocol, its combined score across four representative benchmarks places it at the low-cost inflection point of the observed cost-performance Pareto frontier. Supporting evaluations on tool calling, coding, and instruction following indicate the specialization preserves broad agentic capability. Model weights and partial training data are open-sourced to support research on practical co-work agents and agentic post-training.

Occamy-1.0: Open Pareto-Frontier 35B Intelligence for Co-work

Research area: ML Published: 2026-09-15 arXiv: 2609.11977

Abstract (translated summary)

Co-work agents execute complex workflows that combine information gathering, tool use, coding, and file manipulation across many model invocations. Because cost and latency accumulate over the full episode, their practical value depends not only on peak capability but also on how efficiently that capability is delivered. Yet many steps in everyday work emphasize state tracking, coordination, recovery, and follow-through rather than frontier-scale reasoning.

The authors present Occamy-1.0, a cost-efficient co-work model obtained by further training the post-trained Qwen3.6-35B-A3B checkpoint. Their approach:

Constructing execution-grounded data and environments
Capturing replayable long-horizon trajectories across multiple harnesses
Using staged post-training to develop and consolidate complementary execution capabilities

Results

On a broad range of co-work benchmarks, Occamy-1.0 consistently ranks among the strongest models of its size class, and remains competitive with much larger frontier systems on several tasks.
Under the paper's prescribed evaluation and pricing protocol, its combined performance across four representative benchmarks places it at the low-cost inflection point of the observed cost-performance Pareto frontier.
Supporting evaluations on tool calling, coding, and instruction following show that this specialization preserves broad agentic capabilities.

Open Source

The model weights and part of the training data are open-sourced to support research on practical co-work agents and agentic post-training.

---

*Originally posted on zhichai.net, auto-collected 2026-09-15.*