42 lines
952 B
Markdown
42 lines
952 B
Markdown
<div align="center">
|
||
|
||
# Inference-X Roadmap
|
||
|
||
**Public feature roadmap — vote, contribute, track**
|
||
|
||
[inference-x.com](https://inference-x.com)
|
||
|
||
</div>
|
||
|
||
---
|
||
|
||
## Current Focus (Q1 2026)
|
||
|
||
- [ ] WebGPU backend (browser-native inference)
|
||
- [ ] LoRA adapter support
|
||
- [ ] Structured output (JSON schema enforcement)
|
||
- [ ] Model caching / lazy loading
|
||
|
||
## Upcoming (Q2 2026)
|
||
|
||
- [ ] ONNX Runtime backend
|
||
- [ ] Speculative decoding (2-4× speed improvement)
|
||
- [ ] Multi-modal (vision) unification
|
||
- [ ] ix-scout public map launch
|
||
|
||
## Completed
|
||
|
||
- [x] 19 hardware backends (CUDA, Metal, Vulkan, ROCm, CPU...)
|
||
- [x] OpenAI API compatibility
|
||
- [x] ARM / RISC-V support
|
||
- [x] ECHO multi-agent architecture (research)
|
||
- [x] Organ Architecture (model surgery)
|
||
|
||
## How to vote / contribute
|
||
|
||
Open an issue or comment on existing ones. Items with the most community interest get prioritized.
|
||
|
||
---
|
||
|
||
*Last updated: Feb 2026 · [inference-x.com](https://inference-x.com)*
|