PhiBench
Phi-4-Reasoning-plus 74.2% leads since Apr 2025 Dropped Models answer original prompts that Microsoft's Phi team wrote to test a wide range of skills and reasoning.
In-house: Microsoft never released the tasks, and the Phi-4-reasoning report uses a newer version, PhiBench 2.21.
- Level
- University Undergraduates and entry-level professionals
- Scored
- Graded by a model A language model scores each response, usually against a rubric.
- Answers
- Open-ended Free-form text, documents, designs or media judged as a whole.
Phi-4-Reasoning-plus leads with 74.2% among models released by Apr 2025
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
The newest model with a score here came out in Apr 2025, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6 Astra
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- and 16 more
- 1 Phi-4-Reasoning-plus 74.2%
- 2 Phi-4-reasoning 70.6%
- 3 Phi-4 56.2%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Phi-4-Reasoning-plus | 74.2% | Apr 2025 | Model card | |
| 2 | Phi-4-reasoning | 70.6% | Apr 2025 | Model card | |
| 3 | Phi-4 | 56.2% | Dec 2024 | Model card |
2 models have held number one since its release in Dec 2024
Lab claims count from the day the card came out, independent runs from the model’s release.
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
No launch post has listed it since Apr 2025
Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.
| Date | Launch post | Score | As the post writes it |
|---|---|---|---|
| 30 Apr 2025 | Phi-4-reasoning | 70.6 | PhiBench 2.21 |
| 30 Apr 2025 | Phi-4-Reasoning-plus | 74.2 | PhiBench 2.21 |
| 12 Dec 2024 | Phi-4 | 56.2 | PhiBench (internal) |
3 system cards report it, from 1 lab
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 30 Apr 2025 | Phi-4-reasoning technical report | 70.6 | PhiBench 2.21 (Table 2) | Capabilities |
| 30 Apr 2025 | Phi-4-Reasoning-plus technical report | 74.2 | PhiBench 2.21 (Table 2) | Capabilities |
| 12 Dec 2024 | Phi-4 technical report | 56.2 | PhiBench (internal) (Table 1) | Capabilities |
More in Other
- Personal Superintelligence Index (Meta) In use Meta's composite of seven personal and knowledge-work evals: Agents' Last Exam, AutomationBench-AA, FocusBench, JobBench, GDP.pdf, VBT and DRACO.
- Artificial Analysis Intelligence Index v5 In use Version 5 of Artificial Analysis's index: nine evaluations in four equal categories, most with private test sets, run three times each in its own agent harness.
Gemini 4 Carbon 59.0 - Anthropic ECI (AECI) Only in system cards Index that combines a basket of capability benchmarks into one capability score to track trends across Anthropic models. Claude Opus 5.5 169.4
- Qwen in-house English automatic evaluation Only in system cards Models answer Qwen in-house English datasets on knowledge, comprehension, coding, math and related skills. Skills and weights change between reports.
Qwen2-72B 72.9% - Recognition of model-written user turns (Anthropic) Only in system cards Models write synthetic user turns for real Claude Code transcripts, then judge which version is real. Score is accuracy above chance. Claude Mythos Preview 0.35
- LiveBench Dropped Models answer monthly refreshed questions across math, coding, reasoning, language, instruction following, and data analysis. Scored automatically against objective ground truth. Claude Fable 5.1 85.7%