Solvency console
Sign in to open your console
Budgets, actuals and seat right-sizing for your AI spend. Free while we build with design partners.
Welcome to your console
Load a usage export in Actuals, or start with the sample workspace: eight seats, four teams, budgets and alerts already set.
Metered-equivalent
$0
this period
Seat fees
$0
0 seats
Net vs API
$0
positive means seats are the cheap path
Spend over time
Spend by team
assign seats →Alert feed
computed from your data, prices verified 2026-08-26OpenRouter, OpenAI platform, or the generic seats CSV. Claude Code transcripts convert via the CLI in Integrations. Parsed in your browser; usage never uploads.
| Seat | Team | Calls | Metered-equiv. | Net | Top models | Verdict |
|---|
Method: subscription_usage_repriced. Cache reads at the cached-input rate, cache writes at the uncached input rate, unknown models excluded and listed, never guessed.
Build plans
+ New plan in the ComposerThe Composer prices whole stacks (orchestrator, workers, harness, access path) from measured pass rates and saves plans to this account.
Ranked cost per solved task
open full page →Which model is cheapest for what you actually ship?
I want to ship a .
Switches to a tasks-per-month slider for pricing an ongoing volume of work instead of a single project.
DeepSeek V4 Pro ships it for $10 against Claude Fable 5 at a pricier $837 — over ~48 tasks.
~48 tasks: median of 11 shipped mobile apps we measured (middle half 22–113) · how we counted →
SOLVENCY BENCH · LONG-HORIZON · Solvency ran these itself: 30 hard agentic repo tasks, sandboxed tool loop, hidden tests (bench/ in the repo). Own population — the tier models are judged on.
SOLVENCY BENCH · LONG-HORIZON · Solvency ran these itself: 30 hard agentic repo tasks, sandboxed tool loop, hidden tests (bench/ in the repo). Own population — the tier models are judged on.
Priced, awaiting measurement (292) · verified prices for every listed model; cost per solved task stays missing until a benchmark measures it — shown in full, never estimated.
Source: Artificial Analysis (artificialanalysis.ai) · verified 2026-08-21 · Sources
Keeps this scenario URL in your free account.
Assumptions (4) · move modeled rows only
Task tier: multi-file refactor, feature with tests · Tier: moderate; measured rows unchanged.
These move modeled rows only. A model with no published cached-input price is computed uncached and says so in its row, rather than dropping out of its group. Measured rows carry a cost the benchmark observed and cannot be changed by any assumption. Frontier efficiency has no published source and is applied only to modeled rows — untick it to see unadjusted results. Methodology.
Local rig vs cloud
The Local Rig Composer prices Apple Silicon and used-GPU builds against cloud rates: capex vs opex with power costs, fail closed on unprofiled part mixes.
Open the Local Rig Composer →Seat economics report
A standalone file of the current Actuals view: fees vs metered-equivalent, verdicts per seat. Built in your browser; share it with whoever owns the budget.
Load usage data first.
Sample report
Real generator output on the synthetic eight-seat team, the same file design partners receive.
View sampleSeat assignments
rollups feed Overview and BudgetsLoad usage data in Actuals to see seats.
| Seat | Metered-equiv. | Team |
|---|
Team rollup
Monthly budgets by team
Assign seats to teams first, then set budgets here.
Alert bands: 80% of budget raises a warning, 100% raises an overrun. Email and Slack delivery are in design partnership; the feed on Overview is live now.
Model allowlist
Only approved models charge to team budgets; anything else lands in review.
Design partner previewTier ceilings
Cap frontier-tier usage per team; route light tasks to value-tier models.
Design partner previewSeat class rules
Auto-flag seats whose usage fits a cheaper class two months running.
Design partner previewPolicies execute where the spend happens, so enforcement ships with the first design partners rather than ahead of them. Get on the list.
OpenRouter activity export
Drop the CSV from your OpenRouter activity page straight into Actuals.
Upload in ActualsOpenAI platform usage export
The platform usage CSV parses directly; seats come from API key names.
Upload in ActualsClaude Code transcripts
One command turns local session transcripts into the seats CSV.
node tools/seats-report/import-claude-code.mjs --seat you=~/.claude/projects --out usage.csvAnthropic admin usage
Org-wide seat usage synced from the Anthropic admin console.
Request early accessWorkspace
Plan: Design partner · free while the suite is in design partnership.
Data
Everything in this console lives in this browser. Nothing has been uploaded anywhere.