Benchmark avg
51.0
17 tracked scores
Microsoft · ProprietaryPhi-4
Active
Model portfolio
Phi-4
Tap a star to rate
Microsoft · Released 2024-12-11 · v2024-12-11 · Closed · Context 128K tokens
Language Cap
66
Cadle Language Cap 66 · 20%
Benchmark average
—
Not enough benchmark results yet
Pricing
$0.13
$0.50 / 1M out · 160.0 pts/$
MODEL MATRIX · TRUTH CENTER CELLS
Overall
—
Intelligence
100.0
Coding
—
Agentic
—
MMLU
—
GPQA
—
Math
—
LiveCodeBench
—
HumanEval
—
Reason
—
ARC-AGI
—
Coverage AVG
—
Cadle Cap ranks models with within-benchmark percentiles so ARC-AGI ~40 and MMLU ~90 never compete raw. Coverage AVG is an importance-weighted mean of filled expert cells only — blanks are skipped, not zeroed.
CADLE INSIGHTS
Only measured signals
Strongest signal
epoch-eci
Score 130.5
Blended $/1M
$0.63
In $0.13 · Out $0.50
Value index
81.6
Quality ÷ blended $/1M (higher is better)
Context
128K
Provider context window (not max output)
Intel index
100.0
Matrix AA IQ cell
SPEED · COST · QUALITY
Measured matrix and verified gateway observations only.
- Cadle Intelligence
- 100.0
- Input $/1M
- $0.13
- Output $/1M
- $0.50
- Value pts/$
- 160.00
DESCRIPTION + DETAILS
PROVIDER
Microsoft
RELEASE
2024-12-11
CONTEXT
128K tokens
STATUS
ACTIVE
COMMUNITY REVIEWS
Community Reviews & Field Notes
Verified EngineersReal operator evaluations, prompt robustness observations, and deployment notes for Phi-4.
4.5(2)
Leave a Field Observation
Rating:5 - Frontier benchmark
Role:
Tag topic:
Reviews are publicly visible to the engineering community.
Inference OperatorProduction Deployer Verified
Oct 27, 2026
Tested Phi-4 on high-concurrency tool loops. Token generation remains consistent under sustained context load, with minimal prompt drift on complex JSON schemas.
Autonomous Agent LeadEval Maintainer Verified
Oct 23, 2026
Evaluated Phi-4 on multi-turn code synthesis. Code reasoning and diff application succeed on ~88% of test suites. Best paired with strict system prompt guardrails.
MODEL COMPARISON
| Metric | |
|---|---|
| Provider | Microsoft |
| Release | 2024-12-11 |
| Input / 1M | $0.13 |
| Output / 1M | $0.50 |
| Context | 128K tokens |
| Cadle Cap | 66 |
| Cap board | Cadle Language Cap 66 |
| mmlu-pro | 47.6 |
| musr | 23.8 |
| gpqa-diamond | 20.8 |
| math-level-5 | 29.9 |
| bbh | 52.4 |
| ifeval | 5.9 |
| training-compute-flops | 24.0 |
| arc-ai2 | 91.2 |
| mmlu | 84.8 |
| lech-mazur-writing | 62.6 |
| balrog | 11.6 |
| otis-mock-aime-2024-2025 | 13.7 |
DATA PROVENANCE
Scores may use different scales — check the Benchmarks tab for source links.
- Supabase models
- Cadle public observations
SOURCE TRACKING
Evidence linked to this model — listing, hub, benchmarks, or live gateway prices.
API
Microsoft listing
MEASURED CELLS
Same Truth Center numbers as the datasheet above. Accuracy, reliability, safety, and cost scores are not invented.
AA IQ100