Phi 4
ModelOpen sourceActiveby Microsoft · family “phi”
[Microsoft Research](/microsoft) Phi-4 is designed to perform well in complex reasoning tasks and can operate efficiently in situations with limited memory or where quick responses are needed. At 14 billion... (description from the OpenRouter listing)
in: textout: textStructured output
Epoch Capabilities Index
130.4 #151 of 268
90% CI 124.5 – 132.3
Released
12 Dec 2024
Context window
16K tokens
Max output
15K tokens
Input $ / 1M tokens
$0.070
Output $ / 1M tokens
$0.14
Cached input $ / 1M
Unknown
Each benchmark shown separately with its own source
Benchmark results (7)
| Benchmark | Domain | Score | vs best recorded | Setting | Run | Source |
|---|---|---|---|---|---|---|
| Chess Puzzles | games | 1.0% ±1.0 | 1% | — | 28 Aug 2026 | Eval log ↗ |
| Lech Mazur Writing | other | 62.6% | 73% | — | — | External ↗ |
| OTIS Mock AIME 2024-2025 | math | 13.8% ±3.7 | 14% | — | 25 Feb 2025 | Eval log ↗ |
| Balrog | games | 11.6% | 17% | — | — | External ↗ |
| GPQA diamond | science | 56.1% ±2.6 | 59% | — | 31 Jan 2025 | Eval log ↗ |
| MATH level 5 | math | 64.9% ±1.1 | 66% | — | 31 Jan 2025 | Epoch ↗ |
| MMLU | knowledge | 84.8% | 96% | — | — | External ↗ |
Source: Epoch AI Benchmarking Hub (CC BY 4.0). “External” rows are leaderboard results Epoch collects from third parties. Best reported setting per benchmark is shown.
API list price over time
Price history
$ per 1M tokens
13 observations (OpenRouter listing + Internet Archive snapshots). History accumulates with every ingest run; a single point means no change has been observed yet.
Events
- Benchmark32d agoEpoch AI evaluates Phi 4Epoch AI Benchmarking Hub