GPU-accelerated ALM platform for insurance carriers

Instant actuarial results on hardware you own, alongside the systems you already run.

Valuations, reserves and full ALM projections in seconds to minutes on a workstation GPU. Nested stochastic VM-21 in minutes, not days. Your data formats, your network, your operating model — nothing to rip out.

Why the name InstaVal?

Instant, five ways.

Instant results
Valuation and reserve answers in seconds to minutes on production-size blocks, on a GPU you already own. Fast enough to iterate — tweak, rerun, compare — before the meeting ends. And any past result re-derives from its inputs in seconds, so nothing has to be warehoused and nothing can drift.
Instant ready
Try it in the browser today, or download one self-contained bundle. Your CSVs load as-is — no modeling language, first valuation the same day. Licensing is a signed credit token: buy credits, paste the token, run. Online or air-gapped, same file, no procurement cycle.
Instant scale
Same binary on a laptop, a workstation, or a rented 64-core server. No cluster to provision. A million-record ALM book at about two minutes a scenario on a ~$1K NVIDIA card.
Instant enhancements
Eight releases between June and September 2026. New product types, solvers and speed ship as point releases that upgrade in place — not annual upgrade projects.
Instant audit trail
Every run records its inputs, assumptions and version. Any number on any report re-derives on demand — full traceability without a results warehouse.

Every one of these is measured or shipping today. The rest of this page is the evidence.

The opportunity

Keep your infrastructure. Lose the wait.

Most carriers run an actuarial engine plus a warehouse, ETL, BI and reconciliation layer built around batch runtimes. InstaVal drops in beside all of it — nothing has to change on day one.

Works with what you have
  • CSV, Parquet or Excel in — your existing extracts, as-is.
  • Results land query-ready in compact columnar files your warehouse and BI tools already read.
  • Run in parallel with your current engine; migrate product by product, or not at all.
  • Your hardware, your network, your operating model.
When reruns take seconds
  • Stored results become a convenience, not a system of record you have to defend.
  • Reconciliation becomes a button press — regenerate any number from its versioned inputs.
  • Retention shrinks to megabytes of inputs plus the results you choose to keep.
  • Per-contract audit trace — cash flows, states, reserves — in milliseconds.

The machinery around the engine can shrink at your pace, as far as you want — the savings are real, but they are your call, on your timeline.

Coverage

A complete ALM platform.

Assets and liabilities projected together on one engine, with the statutory frameworks and strategy solvers built in.

Assets
  • Bonds
  • Mortgages
  • MBS
  • CMO (9 tranche types)
  • Equities
  • Cash & equivalents
  • Derivatives (6 types)
  • External assets
Liabilities
  • Term life
  • Universal life + ULSG
  • Indexed UL
  • Variable UL
  • Non-par whole life
  • Disability income
  • Long-term care
  • Fixed annuities (CARVM)
  • SPIA / DIA
  • Variable annuities
  • Pensions (DB · PRT)
  • Reinsurance (YRT + coinsurance)
  • External liabilities
Regulatory / risk
  • VM-20 PBR (UL / IUL / VUL)
  • VM-21 CTE(70)
  • AG 36 (IUL reserves)
  • NAIC Life RBC (LR031)
Strategy solvers
  • Defeasance solver
  • Capital funding solver

Strategy solvers run root-finds over the full GPU projection engine: Defeasance delivers the market-consistent measurement that enables IFRS 17 Phase 2; Capital Funding solves funding IRR and required capital at a hurdle rate.

The core

One generalized Markov transition engine.

Every life and health product runs on the same mathematical core: define the states a contract can be in and the transition probabilities between them, and the engine propagates the full state distribution every period. Term, whole life, UL / IUL / VUL, DI and LTC are one linear-algebra formulation — a new product is a new state model, not a new engine.

Correct

Exact distribution propagation, not point-estimate decrements — competing risks handled properly every period, by construction.

Fast

Matrix math is exactly what GPUs are built for — thousands of contracts' transitions propagate in parallel. The general formulation is the fast one.

Adoption

Easy to set up, natural to use.

CSV in, results out
Standard CSV, Parquet or Excel — no proprietary formats, no model-building language to learn. Your existing data works as-is.
Browser UI
Hosted or installed once for the company, every actuary works from the browser — no per-seat desktop software. Upload, run, drill in; results in CSV, Parquet or Excel.
Parallel run
Run alongside your current system, compare side by side, migrate at your own pace. No big-bang cutover required.

If your admin system can export a CSV, you're ready to run — most carriers load their first portfolio the same day. Conversion isn't a cliff, it's a gradual ramp: start with one product type, validate against your current system, expand.

Governance

Stored results can't drift from the truth.

Legacy model
  • Stored results are the truth.
  • Stale the moment assumptions change.
  • Reconciliation by hand, by a team.
  • “Which version of truth is this?”
InstaVal model
  • Versioned inputs are the truth.
  • Stored results: a convenience, regenerable.
  • Reconciliation = rerun and compare, in seconds.
  • “Click the button again.”

The audit trail isn't a defended artifact. It's a reproducible computation — keep what's convenient, regenerate what's questioned.

Audit & compliance

Full traceability, a fraction of the storage.

Sequence IDs

Every run is identified, versioned and reproducible. Same inputs, same outputs — guaranteed.

Per-contract drill-down

Any single contract: per-period cash flows, state distributions, 3-basis reserves. In milliseconds.

Assumption audit workbooks

Multi-sheet Excel export: policy features, state model, transition matrices, scheduled amounts, projection results, discount rates.

Deterministic reruns

An auditor asks how you got a number — you rerun the exact calculation in front of them.

Sequences

Sensitivity analysis, generalized.

A sequence is a versioned bundle of inputs — portfolio, assumptions, scenarios, parameters. Vary any input across a sequence and the platform runs the sweep: sensitivity analysis on anything, not just the shocks your vendor pre-built.

Easy

Shock any assumption, rate or parameter — no special project, no custom code. Define the sweep, click run.

Controlled

Every leg has a sequence ID and versioned inputs. Reproduce any cell of the sensitivity grid, on demand.

Fast

Each leg runs in seconds to minutes, so a 20-leg sensitivity grid is an hour at most — not a weekend of batch scheduling. When a leg is cheap, you stop rationing questions.

Regulatory

Built-in compliance frameworks.

VM-20 PBR
NPR + DR + SR (CTE70) + exclusion tests for UL, IUL and VUL.
VM-21 CTE(70)
Standard + 50 stochastic scenarios. GMDB (ROP / rollup / ratchet) + GMIB.
AG 36
IUL Actuarial Guideline 36 — Black-Scholes implied rate, 0.03 bps accuracy.
NAIC Life RBC
LR031 covariance formula, C-0 through C-4b, TAC / ACL / ratio.
IFRS 17New
Market-consistent measurement via the Defeasance solver: implied credit spreads and matched-portfolio valuation.
VM-22Roadmap
Non-variable annuity PBR — on the roadmap, timed to the final guidance and phase-in of the new requirements.

US statutory modules are calibrated against published NAIC workbooks with documented error tolerances.

The numbers

From overnight to interactive.

Legacy engines run overnight — or over a weekend — and you review the results tomorrow. InstaVal runs in the same sitting: run, look, change an assumption, run again.
Seconds to minutes
Valuation & reserves on production blocks
500K–2.5M records
~2 min
Per scenario, 1M-record ALM book
Steady state, ~$1K NVIDIA card
Minutes, not days
Nested stochastic VM-21 ASPA
On one workstation GPU

Scale — September 2026

A million-record balance sheet on a workstation.

Full ALM book, closed / production mode, 14 product classes (79% liabilities) — on an owned workstation with a ~$1K NVIDIA card, not a datacenter.

1,012,002
Records in one ALM book
~115 s
Per scenario, steady state
8 warm cycles, 112.7–116.4 s

Cold vs. steady: the first scenario after boot is ~9 min (526 s, measured), then ~2 min each — the API server pays warm-up once. The current release pre-warms the request path: first answer on a cold macOS server went from 10.6 s to 2.5 s.

No GPU? The valuation / reserve API scales with cores — 27× at 32 cores, 46× at 64 on a rented x86 server, CPU only.

Deployment

Hardware tiers.

Mac Studio
Per-carrier default
  • 80 GPU cores · up to 512 GB unified memory
  • Daily workloads in seconds to minutes
  • Silent, desk-side, no IT rack
  • The full book runs on a MacBook Air too
NVIDIA GPU
Any modern workstation
  • RTX 4070 Ti class · ~$1K
  • 1M-record ALM at ~115 s / scenario
  • IUL at Pathwise scale in 1.4 s
  • Datacenter GPUs measured at parity
x86 server
CPU-only · owned or rented
  • 32–64 cores, any cloud, by the hour
  • Valuation / reserve API: 46× at 64 cores
  • 100K LTC val-date in 0.86 s
  • ALM runs; the GPU is the ALM lever
Nested stochastic VM-21 ASPA — the hardest problem, on the same hardware

Legacy: 3–5 days on a dedicated cluster, quarterly at best. InstaVal: about two minutes on one workstation GPU, on demand; about 25 seconds with model-point compression (10K→2K, 0.16% CTE error). Outer scenarios are independent — it scales near-linearly across GPUs.

Runs on hardware you already own or rent by the hour. No datacenter GPU. No appliance.

Full benchmark table — September 2026, Apple Metal vs NVIDIA CUDA

What is timed: the compute stage of a time-0 valuation — one projection or reserve pass per contract on a single basis, warm, excluding load and output. Not a multi-basis reserve trajectory. Measured Sep 1 2026 on both machines the same day; warm medians, items per second. Same engine, same outputs to the cent.

WorkloadRun typeApple Metal · M1 Max (2022)NVIDIA CUDA · RTX 4070 TiFaster
Bonds (10K)360-mo cash-flow projection, 1 scenario506K/s322K/sMetal 1.6×
Mortgages (10K)360-mo projection, prepay + default212K/s147K/sMetal 1.4×
MBS (5K)360-mo projection, prepay178K/s495K/sCUDA 2.8×
Term life / Markov (2.8K)Single-basis 120-mo state projection255K/s471K/sCUDA 1.8×
Disability income / Markov (5K)Single-basis 120-mo, 6-state Markov84K/s116K/sCUDA 1.4×
Long-term care / Markov (5K)Single-basis 120-mo, 7-state Markov80K/s100K/sCUDA 1.2×
Fixed annuity CARVM (10K)30-yr greatest-PV-of-surrender reserve2.7M/s14.9M/sCUDA 5.5×
Variable annuity (50K)360-mo fund projection × 10 scenarios209K/s214K/sparity
ESG — 100 scenariosScenario generation, 360 mo17K/s33K/sCUDA 1.9×
ESG — 1,000 scenariosScenario generation, 360 mo52K/s61K/sCUDA 1.2×
ESG — 5,000 scenariosScenario generation, 360 mo62K/s28K/sMetal 2.2×
VM-21 CTE7051-scenario CTE70 reserve, 30 yr1.1M/s2.2M/sCUDA 2.0×
Nested stochastic ASPA150 nodes: 50 outer × 3 dates × 50 inner139/s562/sCUDA 4.0×
VM-20 UL PBRNPR + DR + SR reserves, 70 yr57K/s150K/sCUDA 2.6×
AG36 IULImplied rate + CRVM reserve, 20 yr18.7M/s18.9M/sparity
Pensions (DB) · 100K participantst=0 valuation · 3-basis projection + reserve trajectoryCPU, 8 threads, M1 Max: 0.92 s valuation · 2.16 s 3-basis trajectoryCPU only

Why this is hard to copy

Why incumbents can't copy this.

Even if your current vendor got 100× faster tomorrow, every carrier's warehouse, ETL, BI and reconciliation stack would still exist — because those systems are built around the assumption that results are artifacts to be stored.

InstaVal isn't merely a faster engine. It is built on the idea that any result can be regenerated on demand — so storage, reconciliation and audit can shrink to whatever you choose to keep. An engine built around batch runs can't offer that choice without rebuilding its data model.

See it yourself

Try it today — hosted or downloaded.

Hosted demo
  • Your own live, isolated instance — in the browser in minutes.
  • Nothing to install, no IT ticket, no procurement.
  • Secure access via an emailed sign-in code.
  • Full UI: load a portfolio, run valuations, drill into contracts.
Request a hosted demo
Download & install
  • Single self-contained bundle for Mac (Metal) or Linux (CUDA).
  • Demo mode runs free, on your hardware, behind your firewall.
  • Your real data never leaves your network.
  • The same binary upgrades in place to a paid licence.

Sample portfolios for every product type are included — run your first valuation in the first five minutes, either way.

Next steps

Pilot program.

1
Map your stack

We map your current infrastructure and identify where on-demand compute creates savings — and where it simply fits beside what you run.

2
Proof of concept

Load your actual portfolio. Run your actual workloads. Measure actual wall-clock times.

3
Production deployment

A single workstation or a GPU VM in your own cloud. Parallel run alongside legacy until confidence is established.

The proof of concept is free. We're confident enough in the numbers to let the product speak for itself.

How licensing works

Pay for runs, not for seats.

InstaVal runs on your own hardware — Mac (Apple Metal) or Linux (NVIDIA CUDA) — as a single compiled binary, air-gap capable. Usage is metered in credits: every valuation or projection consumes credits in proportion to what it computes (policies × scenarios × product complexity), debited before the run and refunded automatically if the engine errors. Nothing leaves your machine except an optional usage record.

Every tier runs the same meter, and a credit is one dollar — the number on your dashboard is the number on your invoice. Standard licences are prepaid: buy credits in the portal whenever you need more, no subscription, no per-seat licence. Enterprise licences are an annual contract with uncapped usage — run as much as you like, with no balance to watch — billed at the frequency you prefer; the meter keeps reporting consumption so the licence can be sized to your book. Terms are scoped in a conversation, not on a price list.

Try it first

Hosted demo instances for prospects, and a UI-only download tier that needs no licence at all.

Buy only what you use

Metered per run, fractional credits, no ceiling and no minimum term on Standard.

Stay sovereign

Air-gapped installs replenish by pasting a signed token. Nothing phones home unless you let it.

A vendor-hosted cloud tier is on the roadmap and will be built when customer demand warrants it. Today, the same binary runs on a GPU VM you rent in your own cloud account — on-prem under the licence, with none of the idle cost.

Why us

Built from the inside.

InstaVal was built by someone who spent years on the wrong side of this problem — helping carriers navigate their data frustrations, watching actuarial teams wait hours for results they needed in minutes. Every design decision, from the GPU kernel architecture to the web UI, was made by someone who knows what it feels like to be the person waiting for the batch job to finish.

See it on your own data.

A 30-minute conversation: we walk you through the platform, run a live valuation against a portfolio shape that matches yours, and agree what a free proof of concept on your actual book would look like.