Numbers you can check. Methods you can challenge.

FluxMateria benchmarks include accuracy metrics, comparison methodology, and validation datasets.

One engine. Three scientific domains. State-of-the-art accuracy at 3.6 million times the speed of DFT.

FluxMateria holds #1 accuracy on 3 ADMET endpoints, achieves 100% reaction mechanism classification, and delivers sub-1% error across 16 materials properties — with zero fitted parameters, no training data, and no GPU. Every number below is published with full methodology. Reproduce any result yourself.

3
ADMET endpoints
at #1 worldwide
100%
Mechanism classification
336 experimental cases
<1%
Materials error
16 properties validated
178K
Compounds validated
leave-one-out protocol
3.6M×
Faster than DFT
single-threaded, no GPU
0
Fitted parameters
pure first-principles physics

Our benchmark philosophy

📋

Publish the methodology

How we test, what datasets we use, how we measure. No hidden assumptions.

⚖️

Show head-to-head comparisons

Against established tools where possible. Fair comparisons, same test sets.

🎯

Document validation scope

Where predictions are most and least reliable. We tell you the boundaries.

🔄

Enable reproduction

Enough detail that you could replicate. Trust but verify.

Key metrics at a glance

Summary of performance across core capabilities.

Bond Lengths

0.079%error
453 bonds, 64 elements

Single + multiple bonds across p, d, and s-block. Zero fitted parameters.

Bond Energies

0.289%error
908 bonds, 64 elements

Singles, doubles, and triples. 870/906 within 1.0%. Zero fitted parameters.

Throughput

10,000+mol/hr
full property panel

Single-threaded, no GPU. Scales linearly with cores for batch jobs.

ADMET Panel

178Kvalidated
leave-one-out, 8 endpoints

PPB, BBB, solubility, metabolism, permeability, hERG, DILI, CYP. 3 are #1 SOTA.

Materials

<1%MAPE
16 properties, universal engine

Band gap 0.7 eV MAE (1,048 materials). Core holdout 1.2%. Gemstone color 19/19.

Benchmarks by module

Detailed performance data for each capability.

💊 ADMET

Production
80.9%
CYP Panel Accuracy
93.3%
BBB Accuracy
~350
mol/sec
178K
Compounds Validated
  • BBB: 93.3% accuracy (7,807 LOO, v8 Hybrid) ✓
  • Solubility: 0.06 logS MAE (9,982+ LOO, v14 Hybrid — #1 SOTA) ✓
  • CYP Panel: AUPRC 0.798, 80.9% acc (62,794 LOO, v5 Hybrid) ✓
  • Permeability: MAE 0.502, 73.1% acc (41,175 LOO, v1 Hybrid) ✓
  • Metabolism: Spearman 0.692, 82.8% acc (38,576 LOO — #1 SOTA) ✓
  • PPB: 2.24% LOO MAE (14,288 LOO, v49.2 Hybrid — #1 SOTA) ✓
  • hERG: AUROC 0.850 (8,879 LOO, v1 Hybrid) ✓
  • DILI: AUROC 0.878 (907 LOO, v1 Hybrid) ✓

178K compounds validated via LOO across 8 hybrid endpoints. 3 are #1 SOTA.

Full Results → Methodology

🔬 Materials

Production
0.703 eV
Band Gap MAE
1.1668%
Core S2 MAPE
<1%
Universal 16 (strict + OOF)
2.741 ms
Universal strict runtime

Band Gap Benchmark

1,048 materials
Overall MAE0.703 eV Metals (exp = 0)0.285 eV Non-metals (exp > 0)1.032 eV

Core Holdout (5 properties)

Lower MAPE is better
FLUX S2 (family holdout)1.17% FLUX S3 (interaction holdout)1.38% AFLOW S236.1% JARVIS S210.9% Matbench S218.4%

Universal Layer 7 (16 properties)

Strict + out-of-family
All 16 strict<1% All 16 out-of-family<1% Worst OOF scenario0.894% Runtime mean2.7 ms Gemstone color match19/19
What this means: FLUX now has two primary validated tracks: near-1% strict holdout error on core thermo-mechanics (with external apples-to-apples baselines), and sub-1% strict plus out-of-family performance across a 16-property universal runtime path. It also includes a curated mini-benchmark showing defect-context color flexibility for real-time UI exploration.
Universal Benchmark → Band Gap Benchmark Module Page

Battery Electrochemistry

Production
1.0
Family Accuracy
0.149 V
Holdout Voltage MAE
5 / 5
Scenario Alignment
26.8 s
End-to-End Workflow
  • Calibrated holdout benchmark tracks capacity, voltage, transport, cycle, electrolyte, interface, cost, and manufacturing together ✓
  • Energy-dense cobalt-free screen lifts LiMnO2 to the top ✓
  • High-voltage frontier screen lifts LiNiPO4 to the top ✓
  • Fast-charge and cycle-life screens surface transport- and stability-led families instead of a single default winner ✓
  • The same pipeline yields different leaders for bulk, interface, battery-native, and build questions ✓

This benchmark validates the battery-native decision layer as a screening and prototype-handoff engine, not as a replacement for electrochemical lab validation.

Full Results → Case Study Module Page

🧲 Curie Temperature

Production
4.6%
Overall MAPE
−0.03%
Mean Bias
107
Magnetic Materials
17
Material Families
  • 4.6% MAPE across 107 materials from composition only, zero fitted parameters ✓
  • 17 families: ferrites, rare-earth intermetallics, double perovskites, manganites… ✓
  • 89% within 5%, 96% within 10% of experimental Tc ✓
  • Near-zero bias (−0.03%) — no systematic over- or under-prediction ✓
Full Results → Module Page

⚛️ Atomic & Magnetic Properties

Production
2.5%
EN MAPE
1.7%
IE MAPE
84
MM Materials
5
Properties
  • Electronegativity 2.5% MAPE (75 elements), ionization energy 1.7% (27), electron affinity 1.0% (28) ✓
  • Magnetic moment: 100% pass (84/84 materials); metallic intermetallics 3.1% MAPE ✓
  • Saturation magnetization: 100% pass (10/10 materials at ±50% tolerance) ✓
  • All from composition only — zero fitted parameters, zero training data ✓
Full Results → Module Page

📈 Spectroscopy

Production
6.2%
UV-Vis Error
<1%
IR Error
0.3-0.5
NMR MAE (ppm)
50
UV-Vis Molecules
  • UV-Vis: 6.2% mean error, 50 molecules, 6 categories ✓
  • IR: <1% error, 32 NIST molecules validated ✓
  • NMR: 0.3-0.5 ppm MAE, 10 SDBS molecules, 5 nuclei ✓
Full Results → Module Page

⚗️ Mechanism Discovery

100%
Mechanism Accuracy
336/336
Cases Correct
7.4
kJ/mol MAE
1,000,000x
Faster
  • 336/336 experimental test cases (SN1/SN2/E1/E2/E1cb) ✓
  • 10,000 random physical consistency tests ✓
  • Head-to-head comparison with DFT (B3LYP) ✓
  • Every prediction traceable and reproducible ✓
Full Methodology & Results → Module Page

MechanismOS

Production
100%
GOLD Direct Ea
94.86%
SILVER Arrhenius Ea
100%
Experimental-only rows
2 tiers
Official validation gate
  • GOLD: 154/154 direct measured activation barriers passed ✓
  • SILVER: 1255/1323 Arrhenius-derived barrier checks passed ✓
  • Official experimental source provenance documented per benchmark tier ✓
  • Control surfaces, optimizer, and audit evidence pack validated end-to-end ✓
Full Results → Module Page

🧪 Synthesis Planning

Production
3.1%
Barrier MAE
29/29
Reaction Types
200
Specific Reactions
<50ms
Per Plan
  • 29 reaction-type barriers at 3.1% MAE (100% pass rate) ✓
  • 200 specific reactions at <1% MAE (72 exact matches) ✓
  • 15 disconnection SMARTS patterns validated ✓
  • All barriers fully auditable and reproducible ✓
Full Results → Module Page

🔥 Reaction Enthalpy

Production NEW
3.5%
MAPE
157
Reactions Tested
89%
Within 5%
<1ms
Per Reaction
  • 157 reactions from NIST WebBook at 3.5% MAPE, 10.0 kJ/mol MAE ✓
  • 12 categories: combustion, radical, formation, halogen, nitrogen, ozone ✓
  • Hess’s law with 3-tier species lookup + universal bond engine ✓
  • Phase notation: C(s), C(g), H2O(l) — disambiguates reference states ✓
Full Results →

⚡ Electron Transfer

Production
26/26
Tests Pass
2–3×
Tunneling Enhancement
Literature
Decay constant match
~150ms
Per Pair
  • Marcus rate constants with FLUX tunneling corrections ✓
  • Through-bond decay constant matches literature ranges ✓
  • Normal, activationless, and inverted Marcus regimes ✓
  • All coupling deterministic and traceable ✓
Full Results → Module Page

🧪 Solvation

0.3295
MAE (kcal/mol)
642
FreeSolv Cases
4
Native Non-Water Carriers
  • Explicit hydration benchmark: 0.3295 kcal/mol MAE on 642 FreeSolv cases ✓
  • Official packet includes summary JSON, case CSV/JSON, and methodology ✓
  • Water externally benchmarked; methanol, ethanol, acetonitrile, and DMSO tracked ✓
Full Results →

🧬 BioTarget

Beta
0.772
Pearson r (CASF-2016)
91%
MoA Accuracy
1.28
MAE (pKi)
10,065
Targets
  • Binding affinity: Pearson r = 0.772 on CASF-2016 (270 complexes) ✓
  • MoA prediction: 91% accuracy on ChEMBL validation ✓
  • Target identification: AUC 0.980 ✓
  • Selectivity profiling: planned ⏳
Full Results →

⚛ Chemistry

Production
0.079%
Bond Length Error
0.289%
Bond Energy Error
1,361
Total Observables
64
Elements
  • Bond lengths: 453 bonds (391 single + 62 multiple), 0.079% mean error ✓
  • Bond energies: 908 bonds, 0.289% mean error, 870/906 within 1.0% ✓
  • First-principles derivation: zero fitted parameters ✓
  • Coverage: 24 p-block + 30 d-block + 10 s-block elements ✓
Full Results → Module Page

Validation scope

Where FluxMateria predictions are most and least reliable.

We document the boundaries of reliable prediction space:

  • ⚠ Novel chemotypes far from validated chemical space
  • ⚠ Specific endpoints with limited experimental data
  • ⚠ Edge cases identified through validation

Confidence indicators in predictions reflect these boundaries. Low confidence = verify experimentally.

Reproducibility

We want you to verify our claims.

Benchmark datasets and evaluation scripts are available to pilot participants.

Request Pilot Access