Model comparison
Claude Opus 5 vs Phi-4
Claude Opus 5 is the stronger model overall, scoring 67.8 to 31.2 on the Noometry Index. Phi-4 costs 114× less per token, which makes it the better buy when Claude Opus 5's lead doesn't matter for your workload.
Last verified . 22 shared benchmarks.
MicrosoftSummary
- They share 22 benchmarks with published results for both. Claude Opus 5 scores higher in 9 categories and Phi-4 in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Opus 5 leads 86.2 to 20.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 98.9% for Claude Opus 5 and 13.8% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $5 / $25 for Claude Opus 5.
- Claude Opus 5 accepts more context: 1M tokens versus 128K.
- Phi-4 has downloadable open weights; the other is API-only.
Side by side
| Claude Opus 5 | Phi-4 | |
|---|---|---|
| Provider | Anthropic | Microsoft |
| Noometry Index | 67.8 | 31.2 |
| Released | 2026-07-24 | 2024-12-11 |
| Weights | Proprietary | Open |
| Context window | 1M | 128K |
| Max output | 128K | 4K |
| Input $ / M tokens | $5 | $0.07 |
| Output $ / M tokens | $25 | $0.14 |
| Results tracked | 57 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 5 leads
Claude Opus 5: 67.5 (#5), Phi-4: 34.4 (#239)
| Benchmark | Claude Opus 5 | Phi-4 |
|---|---|---|
| LMArena Coding | 1534 | 1231 |
| DeepSWE | 73.6% | — |
| FrontierCode | 53.4% | — |
| CursorBench | 46.6% | — |
| LMArena WebDev | 1691 | — |
| FrontierSWE | 52% | — |
| SciCode | 56.4% | — |
| WeirdML | 91.8% | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
| ALE-Bench | 2,165 | — |
Agentic & Tool Use Claude Opus 5 leads
Claude Opus 5: 55.6 (#1), Phi-4: 22.8 (#128)
| Benchmark | Claude Opus 5 | Phi-4 |
|---|---|---|
| BALROG | 63.4% | 11.6% |
| APEX-Agents | 65.8% | — |
| Berkeley Function Calling Leaderboard | — | 28.8% |
| OSWorld 2.0 | 31.4% | — |
| τ²-bench Banking | 48.7% | — |
| PostTrainBench | 35% | — |
| GBAEval | 79.6% | — |
| GDP.pdf | 24% | — |
| Vending-Bench 2 | 11,182 | — |
Reasoning Claude Opus 5 leads
Claude Opus 5: 77.2 (#4), Phi-4: 17.7 (#291)
| Benchmark | Claude Opus 5 | Phi-4 |
|---|---|---|
| Chess Puzzles | 42% | 1% |
| LMArena Hard Prompts | 1526 | 1220 |
| Epoch Capabilities Index | 162.78 | 130.42 |
| ARC-AGI-2 | 90.4% | — |
| SimpleBench | 80.6% | — |
| NYT Connections (extended) | 94.3% | — |
| ARC-AGI-1 | 97.5% | — |
| CritPt | 29.1% | — |
| EBR-Bench | 45.7% | — |
| LiveBench Reasoning | — | 47.8% |
| Mystery Game Puzzles | 59% | — |
| DTBench | 97.9% | — |
| LiveBench Data Analysis | — | 45.2% |
| LMCA | 64.5% | — |
| Bench to the Future 3 | 0.12 | — |
| LiveBench | — | 41.6% |
Math Claude Opus 5 leads
Claude Opus 5: 86.2 (#8), Phi-4: 20.8 (#285)
| Benchmark | Claude Opus 5 | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 98.9% | 13.8% |
| LMArena Math | 1531 | 1246 |
| FrontierMath (Tiers 1-3) | 85.6% | — |
| FrontierMath Tier 4 | 73.2% | — |
| ProofBench | 99% | — |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |
Knowledge Claude Opus 5 leads
Claude Opus 5: 66.8 (#9), Phi-4: 32.6 (#209)
| Benchmark | Claude Opus 5 | Phi-4 |
|---|---|---|
| GPQA Diamond | 93.9% | 56.1% |
| LMArena Expert | 1557 | 1203 |
| SimpleQA Verified | 59.9% | — |
| Confabulations | — | 29.4% |
| Vectara Hallucination Rate | — | 3.7% |
| MMLU | — | 84.8% |
Multimodal Not comparable
Claude Opus 5: 50.8 (#8), Phi-4: —
| Benchmark | Claude Opus 5 | Phi-4 |
|---|---|---|
| LMArena Vision | 1319 | — |
| Blueprint-Bench 2 | 30.4% | — |
| Furniture Assembly | 60.8% | — |
| LMArena Document | 1516 | — |
Multilingual Claude Opus 5 leads
Claude Opus 5: 58.8 (#4), Phi-4: 37.2 (#237)
| Benchmark | Claude Opus 5 | Phi-4 |
|---|---|---|
| LMArena Non-English | 1501 | 1197 |
| LMArena Chinese | 1574 | 1212 |
| LMArena French | 1519 | 1224 |
| LMArena German | 1524 | 1222 |
| LMArena Japanese | 1516 | 1158 |
| LMArena Korean | 1521 | 1151 |
| LMArena Russian | 1507 | 1209 |
| LMArena Spanish | 1519 | 1234 |
Instruction Following Claude Opus 5 leads
Claude Opus 5: 79.2 (#7), Phi-4: 60.4 (#251)
| Benchmark | Claude Opus 5 | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1517 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
Long Context Claude Opus 5 leads
Claude Opus 5: 46.5 (#21), Phi-4: 36.9 (#226)
| Benchmark | Claude Opus 5 | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1515 | 1217 |
Writing & Preference Claude Opus 5 leads
Claude Opus 5: 79.2 (#1), Phi-4: 40.5 (#244)
| Benchmark | Claude Opus 5 | Phi-4 |
|---|---|---|
| LMArena Text | 1507 | 1217 |
| LMArena Creative Writing | 1491 | 1182 |
| LMArena Multi-Turn | 1499 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| EQ-Bench Creative Writing | 2133 | — |
| EQ-Bench 4 | 1385 | — |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is Claude Opus 5 better than Phi-4?
Claude Opus 5 is the stronger model overall, scoring 67.8 to 31.2 on the Noometry Index. Phi-4 costs 114× less per token, which makes it the better buy when Claude Opus 5's lead doesn't matter for your workload.
Which is cheaper, Claude Opus 5 or Phi-4?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; Claude Opus 5 lists at $5 and $25.
Is Claude Opus 5 or Phi-4 better for coding?
Claude Opus 5 scores higher on coding benchmarks: 67.5 versus 34.4 in the Noometry coding category.
Which has the bigger context window?
Claude Opus 5 does, with 1M tokens against 128K.
How many benchmarks do Claude Opus 5 and Phi-4 share?
22 benchmarks have published results for both models. Claude Opus 5 has 57 scored results on Noometry and Phi-4 has 37.