MODELS
ClawsGO runs on the best models.
We don't train foundation models. We pick the best ones, route them well, and put them on your menu.
| MODEL | BEST FOR | REF. MULTIPLIER |
|---|---|---|
CLAWSGOPost-trained on open-source models, optimized for long-horizon research tasks. | ||
ClawsGO 1.5 Lithium clawsgo-1.5-lithium · Post-trained open model | Fast, everyday research tasks. The default | x0.5 |
ClawsGO 1.5 Silicon clawsgo-1.5-silicon · Post-trained open model | The stronger all-rounder | x1.0 |
ClawsGO 1.5 Titanium clawsgo-1.5-titanium · Post-trained open model | The hardest, longest-horizon tasks | x1.5 |
ANTHROPICSonnet 5.5 and Opus 5.5 as workhorses; the Fable 5.1 flagship, from Ultra | ||
Claude Sonnet 5.5 claude-sonnet-5-5 · Anthropic | The everyday workhorse. Fast and steady | x1.5 |
Claude Opus 5.5 claude-opus-5-5 · Anthropic | Long-horizon analysis and deep reasoning | x3.0 |
Claude Fable 5.1 claude-fable-5-1 · Anthropic | The Mythos-class flagship for the hardest problems. Available from Ultra | x10.0 |
OPENAITerra for speed, Astra for depth | ||
GPT 5.6 Terra gpt-5.6-terra · OpenAI | Fast, disciplined reasoning | x1.0 |
GPT 6 Astra gpt-6-astra · OpenAI | The deepest reasoning | x4.0 |
Fast, everyday research tasks. The default
REF. MULTIPLIER
x0.5The stronger all-rounder
REF. MULTIPLIER
x1.0The hardest, longest-horizon tasks
REF. MULTIPLIER
x1.5The everyday workhorse. Fast and steady
REF. MULTIPLIER
x1.5Long-horizon analysis and deep reasoning
REF. MULTIPLIER
x3.0The Mythos-class flagship for the hardest problems. Available from Ultra
REF. MULTIPLIER
x10.0Fast, disciplined reasoning
REF. MULTIPLIER
x1.0The deepest reasoning
REF. MULTIPLIER
x4.0Reference multipliers are relative to ClawsGO 1.5 Silicon (x1.0): each model has its own rates for input, output and cached tokens, and the multiplier is the approximate cost ratio on a typical task, rounded to 0.5; cached input is heavily discounted. Model access follows your plan: Free includes Lithium, Plus unlocks Sonnet, Terra and Silicon, Pro unlocks Opus, Astra and Titanium, and Ultra unlocks Fable 5.1.
Built on the strongest models.
We don't train foundation models. We put the best ones on your menu.
- Claude Fable 5.1
- Anthropic's Mythos-class flagship. Available from Ultra
- Claude Opus 5.5
- The workhorse for long-horizon analysis and deep reasoning
- GPT 6 Astra
- OpenAI's deepest reasoning
Usage is metered in credits, with each model's reference multiplier on the menu. Compute, execution environments, and storage are included in your plan. You only pay for intelligence.
Why would you do that?
Because we're growing, and this is how we'd rather spend the money.
The models are real. You can audit which model ran each step.
Try it with your hardest problem.
Or pick it yourself.
ClawsGO 1.5 Lithium is the default. Each plan unlocks stronger models, switchable per task. When better models ship, we evaluate them and bring them here.
No credit card · Running in minutes