Synth AI
ResearchWorkshopBlogDocsSign in
ResearchWorkshopBlogDocsSign in

Synth Workshop · v0.9.3

Increase your Experiments Per Minute

Workshop combines fast iteration, interactive data visualization, and durable research infrastructure in one local and API-native workbench. Run evals, debug environments, optimize with GEPA, fine-tune models, and move directly from evidence to the next idea.

Build → run → inspect → improve → run again

Download the macOS appCreate a Workshop accountFirst research workflow

North-star metric

Experiments Per Minute

Access

Local and API, as peers

Platform

macOS 14+ · Apple silicon

Product pillars

Four reasons the loop stays fast.

Speed

Compress the distance between an idea and trustworthy evidence. Setup, execution, inspection, and the next run belong in one loop.

Visualize data

Runs, traces, scores, candidates, and artifacts become interactive while the work is happening—and stay useful afterward.

Core data and algorithm infrastructure

Environments, datasets, evals, GEPA, SFT, traces, and provenance share nouns and contracts so new workflows compose instead of restarting from scratch.

Local first-class, API first-class

Local models, local data, and local execution are peers to ChatGPT login, direct provider APIs, and Synth account-backed services.

Need to exercise interactive software directly? Use CUA as part of the environment QA loop. It supports the research workflow without replacing it.

Two ways in

Start locally, or with Synth Cloud.

No Synth account

Use Workshop locally

Download the macOS app and begin without an account. Local models, local data, and local execution are enough to start a research loop.

Download the macOS app

Optional account-backed access

Use Workshop with Synth Cloud

Create an account, receive included usage, connect the app, and unlock account-backed eval, optimization, SFT, and cloud-model work.

Create a Workshop account

Research loop

Evals, environments, GEPA, and SFT in one workbench.

Evals

Run and compare evaluations with evidence you can inspect, not a score you have to trust blindly.

Environment QA

Build, exercise, and debug environments until their behavior is stable enough to optimize against.

GEPA and optimizers

Improve systems with evaluation-driven search. Workshop consumes Optimizers as an independent service, not a hidden internal.

Supervised fine-tuning

Train, checkpoint, and evaluate with the same evidence loop as evals and GEPA.

Providers

Use the access path you already have.

ChatGPT login

Use the ChatGPT account you already work with. Workshop does not require you to abandon that path.

Local models

Run on-device models without an account. Local inference stays on your Mac.

Direct provider APIs

Bring your own credentials. Provider keys remain in the native app's custody.

Synth API

Account-backed Synth services add managed capability, shared usage, and cloud scale without changing the research loop.

Open source

Workshop is open source and consumes independent infrastructure.

Workshop is Apache-2.0 licensed and includes agent-neutral bootstrap, doctor, and build scripts. Containers and Optimizers remain standalone open-source projects whose contracts Workshop consumes.

Workshop source and build instructionsContainersOptimizersWorkshop docs

Onboarding

Reach a first successful eval, environment QA, GEPA, or SFT loop.

Billing and usage

What stays free and local, what the account enables, and how paid work is governed.

Release log

Every published Workshop line, with checksums and provenance.

Troubleshooting

Install, launch, upgrade, providers, and support for the macOS app.

Checksum and provenanceAll changelog entries

Current published download line: Synth Workshop 0.9.3 (artifact available). v0.9.3 checksum: 6073029fde9ba4c470d5956bf4b6c5d8b26eaa5d81fbf65485899f34b18fe7cd.

© 2026 SynthWorkshopReleasesCookbooksChangelogOpen sourceDocsBook a Demo