Occamy 1.0
Occamy 1.0 is Alibaba Accio’s open-weight co-work model, first described in its September 4 research paper. Further training of Qwen3.6-35B-A3B targets long, stateful workflows using search, code, files, structured APIs, and productivity applications. Its 35B mixture-of-experts backbone activates approximately 3B parameters and retains image input. Full-parameter supervised training, preference optimization, model merging, and sustained-action optimization emphasize recovery and follow-through. Apache-licensed BF16 weights and routine deployment formats are available. The architecture supports 262K context, while supervised training uses 131K sequences; this record uses the latter conservative operating scale.
2026-09-04
35B total, approximately 3B active
Hybrid-attention Mixture-of-Experts Transformer with vision encoder
Apache-2.0
Specifications
- Parameters
- 35B total, approximately 3B active
- Architecture
- Hybrid-attention Mixture-of-Experts Transformer with vision encoder
- License
- Apache-2.0
- Context Window
- 131,072 tokens
- Type
- multimodal
- Modalities
- textimage
Benchmark Scores
Advanced Specifications
- Model Family
- Occamy
- Finetuned From
- Qwen3.6-35B-A3B
- API Access
- Not Available
- Chat Interface
- Not Available
- Variants
- BF16FP8NVFP4GGUFMLXexperimental MTP head
Capabilities & Limitations
- Capabilities
- co-work reasoningtool usecodingvisionlong-horizon task execution
- Known Limitations
- Tool execution and access to files or applications require an external harnessFull reasoning and tool history must be retained across turnsVision modules were frozen during supervised post-trainingExperimental MTP head is separate from validated ordinary decoding
- Notable Use Cases
- productivity agentsrepository maintenancebusiness workflow automation
- Function Calling Support
- Yes
- Tool Use Support
- Yes