Model comparison
Kimi K2 (Jul 2025) vs Phi-4
Kimi K2 (Jul 2025) is the stronger model overall, scoring 41.2 to 31.2 on the Noometry Index. Phi-4 costs 11× less per token, which makes it the better buy when Kimi K2 (Jul 2025)'s lead doesn't matter for your workload.
Last verified . 22 shared benchmarks.
MicrosoftSummary
- They share 22 benchmarks with published results for both. Kimi K2 (Jul 2025) 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 Kimi K2 (Jul 2025) leads 42.7 to 20.8.
- The biggest single-benchmark swing is Berkeley Function Calling Leaderboard: 59.1% for Kimi K2 (Jul 2025) and 28.8% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.57 / $2.30 for Kimi K2 (Jul 2025).
- Kimi K2 (Jul 2025) accepts more context: 262K tokens versus 128K.
Side by side
| Kimi K2 (Jul 2025) | Phi-4 | |
|---|---|---|
| Provider | Moonshot AI | Microsoft |
| Noometry Index | 41.2 | 31.2 |
| Released | 2025-07-12 | 2024-12-11 |
| Weights | Open | Open |
| Context window | 262K | 128K |
| Max output | 262K | 4K |
| Input $ / M tokens | $0.57 | $0.07 |
| Output $ / M tokens | $2.30 | $0.14 |
| Results tracked | 42 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2 (Jul 2025) leads
Kimi K2 (Jul 2025): 42.4 (#102), Phi-4: 34.4 (#239)
| Benchmark | Kimi K2 (Jul 2025) | Phi-4 |
|---|---|---|
| LMArena Coding | 1399 | 1231 |
| SWE-bench Verified (bash only) | 63.4% | — |
| Aider Polyglot | 59.1% | — |
| GSO | 4.9% | — |
| WeirdML | 42.8% | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
| ALE-Bench | 597.5 | — |
Agentic & Tool Use Kimi K2 (Jul 2025) leads
Kimi K2 (Jul 2025): 32.4 (#64), Phi-4: 22.8 (#128)
| Benchmark | Kimi K2 (Jul 2025) | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 59.1% | 28.8% |
| Terminal-Bench | 35.7% | — |
| BALROG | — | 11.6% |
| METR Time Horizons | 59.2% | — |
Reasoning Kimi K2 (Jul 2025) leads
Kimi K2 (Jul 2025): 23.3 (#179), Phi-4: 17.7 (#291)
| Benchmark | Kimi K2 (Jul 2025) | Phi-4 |
|---|---|---|
| LMArena Hard Prompts | 1384 | 1220 |
| Epoch Capabilities Index | 146.01 | 130.42 |
| SimpleBench | 26.3% | — |
| Kagi LLM Benchmark | 64.4% | — |
| Chess Puzzles | — | 1% |
| LiveBench Reasoning | — | 47.8% |
| LiveBench Data Analysis | — | 45.2% |
| ForecastBench | 60.2 | — |
| LiveBench | — | 41.6% |
Math Kimi K2 (Jul 2025) leads
Kimi K2 (Jul 2025): 42.7 (#83), Phi-4: 20.8 (#285)
| Benchmark | Kimi K2 (Jul 2025) | Phi-4 |
|---|---|---|
| LMArena Math | 1397 | 1246 |
| OTIS Mock AIME 2024-2025 | — | 13.8% |
| Omni-MATH | 65.4% | — |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |
| FrontierMath (Feb 2025 set) | 21.4% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
Knowledge Kimi K2 (Jul 2025) leads
Kimi K2 (Jul 2025): 37.3 (#157), Phi-4: 32.6 (#209)
| Benchmark | Kimi K2 (Jul 2025) | Phi-4 |
|---|---|---|
| Confabulations | 20.4% | 29.4% |
| Vectara Hallucination Rate | 17.9% | 3.7% |
| LMArena Expert | 1365 | 1203 |
| GPQA Diamond | — | 56.1% |
| MMLU-Pro | 81.9% | — |
| GPQA (HELM) | 65.3% | — |
| MMLU | — | 84.8% |
Multilingual Kimi K2 (Jul 2025) leads
Kimi K2 (Jul 2025): 49.6 (#130), Phi-4: 37.2 (#237)
| Benchmark | Kimi K2 (Jul 2025) | Phi-4 |
|---|---|---|
| LMArena Non-English | 1372 | 1197 |
| LMArena Chinese | 1415 | 1212 |
| LMArena French | 1379 | 1224 |
| LMArena German | 1387 | 1222 |
| LMArena Japanese | 1349 | 1158 |
| LMArena Korean | 1325 | 1151 |
| LMArena Russian | 1385 | 1209 |
| LMArena Spanish | 1386 | 1234 |
Instruction Following Kimi K2 (Jul 2025) leads
Kimi K2 (Jul 2025): 71.1 (#156), Phi-4: 60.4 (#251)
| Benchmark | Kimi K2 (Jul 2025) | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1348 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
| IFEval | 85% | — |
Long Context Kimi K2 (Jul 2025) leads
Kimi K2 (Jul 2025): 41.2 (#145), Phi-4: 36.9 (#226)
| Benchmark | Kimi K2 (Jul 2025) | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1353 | 1217 |
| Fiction.LiveBench | 66.7% | — |
| CL-bench | 17.6% | — |
Writing & Preference Kimi K2 (Jul 2025) leads
Kimi K2 (Jul 2025): 62.3 (#78), Phi-4: 40.5 (#244)
| Benchmark | Kimi K2 (Jul 2025) | Phi-4 |
|---|---|---|
| LMArena Text | 1380 | 1217 |
| LMArena Creative Writing | 1350 | 1182 |
| Short-Story Creative Writing | 85.6% | 62.6% |
| LMArena Multi-Turn | 1371 | 1206 |
| EQ-Bench Creative Writing | 1666 | — |
| WildBench | 86.2% | — |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is Kimi K2 (Jul 2025) better than Phi-4?
Kimi K2 (Jul 2025) is the stronger model overall, scoring 41.2 to 31.2 on the Noometry Index. Phi-4 costs 11× less per token, which makes it the better buy when Kimi K2 (Jul 2025)'s lead doesn't matter for your workload.
Which is cheaper, Kimi K2 (Jul 2025) or Phi-4?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; Kimi K2 (Jul 2025) lists at $0.57 and $2.30.
Is Kimi K2 (Jul 2025) or Phi-4 better for coding?
Kimi K2 (Jul 2025) scores higher on coding benchmarks: 42.4 versus 34.4 in the Noometry coding category.
Which has the bigger context window?
Kimi K2 (Jul 2025) does, with 262K tokens against 128K.
How many benchmarks do Kimi K2 (Jul 2025) and Phi-4 share?
22 benchmarks have published results for both models. Kimi K2 (Jul 2025) has 42 scored results on Noometry and Phi-4 has 37.