Changelog

New models, faster training and serving, and platform updates from the Freesolo team.

NewServing

Serve models on serve.freesolo.co

Deployed models are reachable on Freesolo's own serving domain, and the Flash CLI points there by default.

ImprovedServing

Cost-optimized GPU selection for every model

Each model now serves on the cheapest GPU that fits it, with per-model LoRA rank and adapter limits.

NewTraining

Validate environments offline with flash env test

Run a fixed set of episodes against your environment.py locally, with no GPU, before you push.

NewTraining

Live per-step training metrics in flash log -f

Follow reward, rollout truncation, and per-step GRPO metrics as a run progresses.

ImprovedTraining

Faster on-policy distillation

Generate-ahead rollout prefetch overlaps generation with teacher scoring to cut idle GPU time.

ImprovedServing

7-10x faster responses on 9B and 27B models

CUDA graphs are enabled for mid-size models, sharply reducing decode latency.

NewTraining

Multi-GPU training jobs

Set gpu.count in your job spec to provision and price multi-GPU pods.

NewTraining

Per-turn or per-episode credit for multi-turn GRPO

Choose how reward is assigned across the turns of a multi-turn rollout.

NewTraining

Named reward metrics per step

Your custom reward metrics now show up per step in the training log.

ImprovedTraining

SFT training up to ~48% faster

Activation shaping, cached tokenization, and chunked loss cut memory use and unlock longer context.

NewPlatform

Live sample completions on the run page

Watch per-step GRPO and OPD sample completions stream in on the run detail page.

ImprovedPlatform

Documentation moved to docs.freesolo.co

Old /docs links now redirect to the new documentation site.

NewTraining

Train on images (multimodal)

Multimodal SFT, single- and multi-turn GRPO, and on-policy distillation for vision models.

NewServing

Multimodal (image) chat serving

Serve vision models with image inputs passed straight through to the model, with correct EXIF orientation.

NewTraining

Chat a specific training step

flash chat RUN_ID/step-N serves that step's exact deployed revision.

FixedTraining

Reliable warm-start from a deployed adapter

Warm-starting from an already-deployed adapter now retries transient throttling instead of failing.

FixedTraining

flash env push uploads your full environment tree

Nested files beyond the root and dataset/ folders are now included.

FixedPlatform

Fixed a render loop in the new training-job composer

NewModels

Qwen3.6-27B is now available

Train and serve Qwen3.6-27B; 9B and 35B serving prices have been refreshed.

NewPlatform

Live run monitoring

Live training metrics, heartbeat progress, and auto-refreshing worker logs on the run detail page.

NewPlatform

Welcome emails and a $50 credit offer

New users get a welcome email and a $50 credit offer after their first Flash run.

FixedTraining

Flash CLI now runs on Windows

Guarded a Unix-only import that previously crashed the CLI on Windows.

FixedTraining

More reliable recovery from GPU provider outages

Interrupted runs now converge to resumed or terminal instead of stranding.

ImprovedServing

32k context on served and trained models

Per-model context was raised to 32k tokens, and the 9B model moved to L40S GPUs.

FixedTraining

Adapter export fixes

Correct key normalization for vanilla PEFT adapters, and clearer Hugging Face Hub error messages.

Curious about post-training? Talk to our team.


Freesolo

Copyright © 2026 Freesolo. All rights reserved.

Freesolo