Transaction: Task risk assessment
An agent operating in this task class commits something outside its own process: money changes hands, a reservation is held, or an order is placed against a third-party system. Unlike drafting or retrieving, a transaction task changes state somewhere the agent does not control, under terms — price, availability, mandate — that were current at the moment of commitment and may not stay current afterward.
Retries are the structural hazard specific to this task class: a transaction system that appears not to have responded may have already succeeded, and applying the same operation again is not a safe default the way re-reading a document is. The four risks tracked below are chosen to reflect exactly that asymmetry between an operation that failed to confirm and one that failed to occur. Think of it as the difference between silence and absence: one of them still moved money.
Which model profile is crossed with a transaction
Before the risk table, the model axis: transaction work crossed against each registry model, with each profile computed from that model's own configuration. A profile is not a credit check and not a reliability rating. It is a fixed characterisation of a configuration, and it is the same characterisation this registry publishes for that model everywhere else.
Sorted, and worth saying plainly: the sort is not a ranking of trustworthiness. 2 of the 10 profiles read transaction work through a dominant house or an element, Gemini 3.7 Flash most strongly at 7 and Qwen3.8-Max least. Across the same 10 rows there are 6 distinct ruling planets, which is the clearest evidence on the page that the right-hand columns are computed rather than assigned.
Rows are computed from registry default parameters — the cutoff shown for each model, temperature 0.700. No amount, currency, or counterparty is an input to any of them, and a certificate uses the parameters submitted with its order.
| Model | Manufacturer | Training cutoff | Ruling planet | Reads this class through | Outlook |
|---|---|---|---|---|---|
| Gemini 3.7 Flash | Google DeepMind | 2025-01 | Venus | dominant house and element (7) | publication pending |
| Grok 4.6 | xAI | 2026-02 | Sun | element (3) | publication pending |
| Claude Fable 5 | Anthropic | 2026-01 | Mars | neither — class-neutral (0) | publication pending |
| Claude Opus 5 | Anthropic | 2026-05 | Saturn | neither — class-neutral (0) | publication pending |
| Claude Sonnet 5 | Anthropic | 2026-01 | Venus | neither — class-neutral (0) | publication pending |
| GPT-5.6 Sol | OpenAI | 2026-02 | Mercury | neither — class-neutral (0) | publication pending |
| GPT-5.6 Terra | OpenAI | 2026-02 | Moon | neither — class-neutral (0) | publication pending |
| Kimi K3 | Moonshot AI | 2026-03 | Moon | neither — class-neutral (0) | publication pending |
| DeepSeek V4 Pro | DeepSeek | 2025-12 | Sun | neither — class-neutral (0) | publication pending |
| Qwen3.8-Max | Alibaba Cloud | 2026-03 | Sun | neither — class-neutral (0) | publication pending |
The four tracked transaction risks
Transaction is the class where the registry's severity scale is least abstract. A severity 3 factor here corresponds to money committed twice; a severity 1 factor corresponds to something the operator could not have established beforehand. All four are tracked on every transaction assessment; three are reported on any one of them.
The Registry tracks 4 risk factors for transaction tasks. They are fixed for the class: they do not adapt to a model, to a configuration, or to the wording of a task. An assessment reports 3 of the 4, each with its severity rating and its description, and which 3 are returned is derived from the submitted task's own seed rather than chosen. The factors themselves are part of the assessment and are not published here.
How a transaction assessment is computed
For transaction tasks, three of the four risk factors above are selected per task instance and paired with the model's fixed profile to form the task-level portion of a full assessment. /method documents exactly how that pairing is computed, byte for byte.
Questions about transaction assessment
- What counts as a transaction task for this registry?
- Any task that commits money, a reservation, or a binding order — a purchase, a booking, a payment, a submitted order — falls under this task class, including one that only initiates the commitment and leaves settlement to another system.
- How many risk factors does an assessment report?
- The registry tracks 4 risk factors for this class and an assessment reports 3 of them, each with its severity rating and its description. Which 3 are returned is derived from the submitted task's own seed. The factors are part of the assessment and are not published on this page.
- Does a transaction assessment rate the counterparty's reliability?
- No. The 10 profiles and the 4 tracked factors on this page describe task-side risk only. The registry does not evaluate, score, or rate the vendor, platform, or counterparty on the other side of a transaction, and one of the 4 exists precisely to mark that boundary.
Order a transaction task assessment
A Standard Assessment costs EUR 1.90, an Extended Assessment EUR 4.90, and a Full Chart Assessment EUR 14.90 — the last of which carries the permanent chart itself. Assessments from EUR 1.90. Machine-readable pricing is published at /pricing.json, with no authentication required and no difference from the figures printed here.
The task class is selected on the order form. The verdict and the recommended execution window are computed from that selection together with the submitted parameters, and they appear on the certificate — not on this page, and not anywhere else in the directory.