NeMo RL v0.8.0 Roadmap

[anwithk]

NeMo RL v0.8.0 Roadmap

ETA: 9/30/2026

This is a community-facing snapshot of the work targeted for the NeMo RL v0.8.0 release. Scope and status may evolve as implementation progresses.

---

1. Training - Model support

Feature	Status	Link
GLM-5.2 support for AutoModel and Megatron Core	WIP	#3172
DeepSeek V4 support for AutoModel	WIP	#2331
Nemotron Nano v3.5 support for AutoModel	WIP	PR #3383
MiniMax M3 support for AutoModel	WIP	#3329
Ultra support	Planned	—

---

2. Low-precision training and quantization

Feature	Status	Link
MXFP8 convergence and benchmark milestones	WIP	#3431
MXFP4 / NVFP4 recipe support	WIP	#3432
Per-token FP8 and NVFP4 quantization	Planned	—

---

3. Inference - SGLang backend

Feature	Status	Link
SGLang support with Megatron Core	WIP	PR #3190
Non-colocated SGLang with Megatron refit, MXFP8, and fault tolerance	WIP	PRs #3190, #3188, #3187, #3189

---

4. Algorithms and data

Feature	Status	Link
PPO with non-colocated generation	WIP	PR #3262
Context compaction	Planned	—
Single-controller asynchronous GRPO	WIP	#2625
Diffusion RL	WIP	#3233
Cross-tokenizer distillation	WIP	#1827, draft PR #3237

---

5. Platform support

Feature	Status	Link
NeMo RL container support and validation for GB300	WIP	#3492