Released: 2025-03-22

Microsoft-Decision-1

Microsoft single-pass decision and classification model outputting calibrated confidence scores without generative token overhead, hosted on Microsoft Foundry and OpenRouter.

Intelligence Index
86.8/ 100
Calibrated multi-domain composite
Throughput Speed
240tok/sec
Output streaming throughput
Time To First Token
130ms
Initial chunk server latency
LiveBench Contam-Free
77.2%
September 2026 suite score

Sourced Benchmark Evaluations (61)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
ARC-AGI-3Active56.9%
93
2025-04-26independentSource ↗
ARC-AGI-2Nearing Saturation68.3%
75
2025-04-19independentSource ↗
FrontierMathActive23.9%
98
2025-04-16independentSource ↗
PutnamBenchActive44.5%
90
2025-04-15independentSource ↗
SWE-LancerDeprecated55.1%
20
2025-04-13independentSource ↗
Omni-MATHDeprecated63.3%
10
2025-04-13independentSource ↗
SoohakActive65%
98
2025-04-11independentSource ↗
FrontierCodeActive47%
98
2025-04-11independentSource ↗
Humanity's Last ExamActive38.5%
93
2025-04-09independentSource ↗
StructFlowBenchNearing Saturation67.9%
75
2025-04-09independentSource ↗
AIMENearing Saturation81.2%
75
2025-04-09independentSource ↗
OlympiadBenchActive70.1%
80
2025-04-08independentSource ↗
MathArenaActive68.4%
99
2025-04-07independentSource ↗
FireBenchActive67.1%
95
2025-04-07independentSource ↗
Terminal-BenchActive56.7%
80
2025-04-07independentSource ↗
OpenAI-MRCRActive72.9%
80
2025-04-06independentSource ↗
SimpleQAActive45.1%
90
2025-04-06independentSource ↗
CodeContestsDeprecated52.7%
10
2025-04-06independentSource ↗
ComplexBenchNearing Saturation65.5%
80
2025-04-05independentSource ↗
SWE-bench ProActive50.2%
90
2025-04-05independentSource ↗
LongBench-ProActive71.2%
80
2025-04-05independentSource ↗
ZeroSCROLLSNearing Saturation69.4%
78
2025-04-04independentSource ↗
MultiPL-EDeprecated74.5%
10
2025-04-04independentSource ↗
LiveCodeBenchSaturated66.4%
39
2025-04-03independentSource ↗
AgentIFNearing Saturation54.3%
80
2025-04-03independentSource ↗
MATHSaturated83.8%
25
2025-04-03independentSource ↗
InfiniteBenchNearing Saturation73.8%
65
2025-04-03independentSource ↗
MercuryDeprecated71.3%
10
2025-04-02independentSource ↗
SCROLLSNearing Saturation69.4%
65
2025-04-02independentSource ↗
MMLU-ProNearing Saturation79.9%
75
2025-04-02independentSource ↗
RULERNearing Saturation79.9%
84
2025-04-01independentSource ↗
TabMWPSaturated78.7%
25
2025-04-01independentSource ↗
CyberSecEvalDeprecated72.9%
10
2025-04-01independentSource ↗
SWE-bench VerifiedSaturated63.2%
25
2025-03-31independentSource ↗
MMLU-ReduxSaturated85.1%
25
2025-03-31independentSource ↗
QuALITYNearing Saturation76.4%
78
2025-03-31independentSource ↗
LongBenchNearing Saturation74.6%
65
2025-03-30independentSource ↗
SWE-benchDeprecated55.1%
10
2025-03-30independentSource ↗
MGSMNearing Saturation81.2%
65
2025-03-30independentSource ↗
AGIEvalSaturated75.3%
25
2025-03-29independentSource ↗
QASPERNearing Saturation72%
65
2025-03-29independentSource ↗
TruthfulQASaturated76.4%
25
2025-03-29independentSource ↗
GSM-HardDeprecated82.1%
10
2025-03-28independentSource ↗
SuperGLUESaturated89.3%
25
2025-03-28independentSource ↗
CodeXGLUEDeprecated76.1%
23
2025-03-27independentSource ↗
MT-BenchNearing Saturation8.2%
75
2025-03-27independentSource ↗
BIG-Bench HardSaturated85.8%
25
2025-03-27independentSource ↗
WinoGrandeSaturated91%
28
2025-03-26independentSource ↗
MBPP+Nearing Saturation82.6%
65
2025-03-26independentSource ↗
IFEvalDeprecated78.3%
25
2025-03-26independentSource ↗
GPQANearing Saturation76.4%
78
2025-03-26independentSource ↗
Needle-in-a-HaystackSaturated97.4%
44
2025-03-25independentSource ↗
HumanEval+Saturated87.5%
25
2025-03-25independentSource ↗
CommonsenseQASaturated91.9%
28
2025-03-25independentSource ↗
MBPPSaturated85.9%
25
2025-03-24independentSource ↗
GSM8KSaturated92.3%
25
2025-03-24independentSource ↗
GLUESaturated94.5%
28
2025-03-24independentSource ↗
MMLUSaturated88.5%
25
2025-03-23independentSource ↗
HumanEvalSaturated92.3%
25
2025-03-23independentSource ↗
ARC-AGISaturated82.3%
25
2025-03-23independentSource ↗
LiveBench (Continuous)Active77.2%
94
2025-02-01independentSource ↗
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run microsoft-decision-1

# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "microsoft-decision-1",
    "messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
  }'

Evaluation & Sourcing Notes

Scores listed for Microsoft-Decision-1 represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.

All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.