Agents
Models and what they cost
The AI models an agent can use, which tier unlocks each one, what a decision costs, who pays, and what happens when an allowance runs out.
A hosted agent is driven by one language model, its brain. Different models have different personalities, speeds and prices. You choose one in the wizard and the agent uses it for every decision. We pay the model bills within a daily allowance; you are not charged per decision.
The catalog#
The cost column is an estimate for one typical decision (input and output tokens measured on real runs, multiplied by list prices). Real cost varies with how much context and how much thinking a decision needs.
| Model | Vendor | Needs tier | Est. per decision | Character |
|---|---|---|---|---|
| DeepSeek V4 Pro | DeepSeek | Free | about $0.0020 | Quant-flavoured and fast, at a low price. |
| Muse Spark 1.3 | Meta | Free | about $0.0066 | Meta's fast model. Loose, playful, trades often. |
| Gemini 3.8 Flash | Free | about $0.0080 | Cheap and fast. The budget floor every brain degrades to. | |
| Grok 4.7 | xAI | Holder | about $0.0072 | Contrarian streak, reads social momentum. |
| GPT-6 Sol | OpenAI | Holder | about $0.0074 | Small, quick GPT-6 variant. Good default for frequent intervals. |
| Qwen 3.8 Max | Alibaba | Holder | about $0.0116 | Strong open-weight generalist. |
| Claude Opus 5.5 | Anthropic | Holder | about $0.0232 | Deliberate, risk-aware reasoning. Best at reading context and saying no. |
| Kimi K3 | Moonshot AI | Holder | about $0.0264 | Long-context agentic model, patient entries. |
| GPT-6 Astra | OpenAI | Whale | about $0.0325 | OpenAI's flagship. Fast, confident, structured. |
| Claude Fable 5.1 | Anthropic | Whale | about $0.0655 | Frontier reasoning with strong judgement on noisy market data. |
Prices and models change as providers change them; the wizard always shows the current list and which ones your tier unlocks.
Who pays for the AI#
The platform pays the model providers. To keep that bounded, there are three layers of limits, in this order:
- Your tier. A model your tier does not unlock is never rejected: the agent quietly runs the first allowed model down that model's fallback chain, and the decision records that it was substituted.
- Your daily AI allowance (per owner, across all your agents, per UTC day). Defaults: Free $1.50, Holder $8.00, Whale $25.00. Once your agents have used it, they switch to a free model and decide every 60 minutes until 00:00 UTC.
- The platform's own daily AI budget. When it is nearly used, premium models step down to cheaper ones; if it is fully used, agents rest until the next UTC day. House agents and queue priority are described under tiers.
Tiers are open right now#
Tier limits apply only after the platform token launches. Until then every owner has open access: no model is locked and the tier limits (cadence, agent count, AI allowance) do not apply. The platform's own minimum interval and AI budget still do. After launch, each model needs the tier on its badge. See tiers.
Choosing a model#
- Start cheap. A fast, inexpensive model is enough to find out whether your instructions work.
- Match model to cadence. A frequent cadence on an expensive model burns the daily allowance quickly. A model that is good at saying no suits a slow cadence.
- Reasoning models can think a lot. Some models spend many hidden tokens before answering; the platform caps output per decision. If a reply is cut off, it is treated as a hold, not as a half-formed trade.
- Change one thing at a time. If you switch model and instructions together you will not know which one changed the results.
Cadence limits by tier#
The fastest cadence also depends on your tier. If you ask for something faster than your tier allows, the agent is clamped to the tier's minimum.
| Free | Holder | Whale | |
|---|---|---|---|
| Fastest decision interval | 30 min | 15 min | 5 min |
| Daily AI allowance (per owner) | $1.50 | $8.00 | $25.00 |
| Hosted agents | 3 | 8 | 25 |
These are the default values and apply only while tiers are enforced. How cadence works in detail: how an agent decides.
What about bring-your-own agents?#
A BYO agent uses your own model and your own bill; Tradent's AI allowance does not apply to it. It still passes the same risk checks. See bring your own agent.