AI Models

Reflects the models in Gydmation's own model picker, as of September 2026

Gydmation can generate workflows using several AI models across three providers. Here's every model available, ranked by accuracy, so you can pick the right one for what you're building.

The accuracy tiers below are Gydmation's own labels — the same ones you'll see in the model picker inside the app — not third-party benchmark scores. They're a practical guide to how much you should double-check a model's output, not a claim about any provider's underlying technology.

Pro vs. Bring Your Own Key

Two of the models below — Claude Opus 5 and Claude Sonnet 5 — are managed models included with the Pro plan. They run on Gydmation's own credits, so there's no API key to set up.

Every other model is available on Bring Your Own Key (BYOK), on any plan including Free: you add your own API key for that provider, and generations run against your key directly rather than drawing from Pro credits. You can switch models from the picker in the composer at any time, including mid-conversation.

All supported models, ranked

Ranked from highest to lowest accuracy. Models tied on accuracy are ordered by how we'd recommend reaching for them.

  1. 1
    Claude Opus 5
    AAnthropic · Pro (managed)
    High accuracy
  2. 2
    Claude Sonnet 5
    AAnthropic · Pro (managed)
    High accuracy
  3. 3
    Claude Fable 5
    AAnthropic · BYOK
    High accuracy
  4. 4
    GPT-5.6 Sol
    OOpenAI-compatible · BYOK
    High accuracy
  5. 5
    GPT-5.6 Terra
    OOpenAI-compatible · BYOK
    Good accuracy
  6. 6
    GPT-4o
    OOpenAI-compatible · BYOK
    Good accuracy
  7. 7
    Claude Haiku 4.5
    AAnthropic · BYOK
    Moderate accuracy
  8. 8
    GPT-5.4 mini
    OOpenAI-compatible · BYOK
    Moderate accuracy
  9. 9
    gpt-oss-120b
    GGroq · BYOK
    Good for testing
  10. 10
    gpt-oss-20b
    GGroq · BYOK
    Good for testing

Model details

1. Claude Opus 5

High accuracy
AAnthropic · Pro (managed) · no API key needed

Our top-ranked model, and one of the two managed models included with Pro. Opus 5 is Anthropic's most capable model, and the difference shows most on the requests that ask a lot at once — long, multi-step automations with several conditional branches, or a workflow built from a single dense prompt.

Best for: your most important or most complex workflows, and anything you'd rather get right the first time than iterate on.

2. Claude Sonnet 5

High accuracy
AAnthropic · Pro (managed) · no API key needed

The second managed model on Pro, and the model we'd point most people to by default. Sonnet 5 sits right behind Opus 5 on accuracy but responds faster, which makes it the better everyday fit rather than a model you save for the hardest requests.

Best for: most day-to-day workflows. If you're on Pro and not sure which model to pick, start here.

3. Claude Fable 5

High accuracy
AAnthropic · BYOK

A high-accuracy Anthropic model available on Bring Your Own Key. If you already hold an Anthropic API key and want Opus/Sonnet-level accuracy without moving to Pro, this is the BYOK option that gets you there.

Best for: BYOK users who want high accuracy and already have an Anthropic key.

4. GPT-5.6 Sol

High accuracy
OOpenAI-compatible provider · BYOK

The highest-accuracy BYOK option if your key is with an OpenAI-compatible provider rather than Anthropic.

Best for: high-accuracy workflows when Claude access isn't the key you have on hand.

5. GPT-5.6 Terra

Good accuracy
OOpenAI-compatible provider · BYOK

A step down from Sol on accuracy, but still solid for most requests — this is the model we reference elsewhere in the docs as a typical BYOK example.

Best for: everyday automations on an OpenAI-compatible key when you don't need the very highest tier.

6. GPT-4o

Good accuracy
OOpenAI-compatible provider · BYOK

A widely used model that many users already hold an API key for. Comparable accuracy to Terra.

Best for: getting started quickly if GPT-4o is the key you already have.

7. Claude Haiku 4.5

Moderate accuracy
AAnthropic · BYOK

Anthropic's fast, lightweight model. It trades some accuracy for speed and cost, so it's worth reviewing its output a bit more closely than Sonnet or Opus.

Best for: short, simple, single-step workflows where speed matters more than handling edge cases.

8. GPT-5.4 mini

Moderate accuracy
OOpenAI-compatible provider · BYOK

The lightweight counterpart to Terra and Sol — fast and inexpensive, with accuracy to match.

Best for: quick drafts you plan to review and adjust yourself, or very simple automations.

9. gpt-oss-120b

Good for testing
Open-weight · served via GGroq · BYOK

An open-weight model served through Groq. It's the larger of the two open-weight options here, so more capable than gpt-oss-20b, though both are labeled for testing rather than production use.

Best for: trying Gydmation out before wiring up a paid API key.

10. gpt-oss-20b

Good for testing
Open-weight · served via GGroq · BYOK

The smaller open-weight model, also served through Groq. Fast to try, but plan to review its output more carefully than any other model on this list.

Best for: a first look at how Gydmation works, before you commit to a key or a Pro plan.

Which model should I use?

Not sure where to start
Use Claude Sonnet 5 on Pro, or Claude Fable 5 on BYOK.
A complex, multi-branch automation, or anything mission-critical
Use Claude Opus 5 — the highest accuracy available.
Just trying Gydmation out, no API key yet
Use gpt-oss-120b or gpt-oss-20b on the Free plan.
You already pay for an OpenAI-compatible key
Use GPT-5.6 Sol for the highest accuracy, or GPT-5.6 Terra / GPT-4o for everyday use.
Simple, single-step workflows where speed and cost matter more than accuracy
Use Claude Haiku 4.5 or GPT-5.4 mini.

Whatever you pick, review a generated workflow before activating it — see our AI Disclaimer for why that matters even on the highest-accuracy models.