Model comparison
GPT-5.4 mini vs Kimi K2 (Jul 2025)
GPT-5.4 mini is the stronger model overall, scoring 44.8 to 40.9 on the Noometry Index. Kimi K2 (Jul 2025) costs 1.7× less per token, which makes it the better buy when GPT-5.4 mini's lead doesn't matter for your workload.
Last verified . 26 shared benchmarks.
Summary
- They share 26 benchmarks with published results for both. GPT-5.4 mini scores higher in 8 categories and Kimi K2 (Jul 2025) in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where GPT-5.4 mini leads 51.4 to 36.9.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 37.9% for GPT-5.4 mini and 64.4% for Kimi K2 (Jul 2025).
- Kimi K2 (Jul 2025) is cheaper at $0.57 / $2.30 per million input/output tokens, against $0.75 / $4.50 for GPT-5.4 mini.
- GPT-5.4 mini accepts more context: 400K tokens versus 262K.
- Kimi K2 (Jul 2025) has downloadable open weights; the other is API-only.
Side by side
| GPT-5.4 mini | Kimi K2 (Jul 2025) | |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 44.8 | 40.9 |
| Released | 2026-03-17 | 2025-07-12 |
| Weights | Proprietary | Open |
| Context window | 400K | 262K |
| Max output | 128K | 262K |
| Input $ / M tokens | $0.75 | $0.57 |
| Output $ / M tokens | $4.50 | $2.30 |
| Results tracked | 46 | 42 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 mini leads
GPT-5.4 mini: 45.0 (#67), Kimi K2 (Jul 2025): 43.1 (#83)
| Benchmark | GPT-5.4 mini | Kimi K2 (Jul 2025) |
|---|---|---|
| WeirdML | 60.3% | 42.8% |
| LMArena Coding | 1438 | 1399 |
| ALE-Bench | 1,189 | 597.5 |
| FrontierCode | 27% | — |
| SWE-bench Verified (bash only) | — | 63.4% |
| Aider Polyglot | — | 59.1% |
| LMArena WebDev | 1397 | — |
| SciCode | 49.9% | — |
| GSO | — | 4.9% |
Agentic & Tool Use Kimi K2 (Jul 2025) leads
GPT-5.4 mini: 29.6 (#74), Kimi K2 (Jul 2025): 32.2 (#59)
| Benchmark | GPT-5.4 mini | Kimi K2 (Jul 2025) |
|---|---|---|
| Terminal-Bench | — | 35.7% |
| Berkeley Function Calling Leaderboard | — | 59.1% |
| DeepResearch Bench | 36.3% | — |
| METR Time Horizons | — | 59.2% |
Reasoning GPT-5.4 mini leads
GPT-5.4 mini: 30.3 (#77), Kimi K2 (Jul 2025): 22.7 (#173)
| Benchmark | GPT-5.4 mini | Kimi K2 (Jul 2025) |
|---|---|---|
| Kagi LLM Benchmark | 37.9% | 64.4% |
| LMArena Hard Prompts | 1424 | 1384 |
| Epoch Capabilities Index | 148.83 | 146.02 |
| ForecastBench | 57 | 60.2 |
| ARC-AGI-2 | 18.9% | — |
| SimpleBench | — | 26.3% |
| NYT Connections (extended) | 61.8% | — |
| ARC-AGI-1 | 63.7% | — |
| CritPt | 10% | — |
| Chess Puzzles | 24% | — |
| Thematic Generalization | 61.7% | — |
| Mystery Game Puzzles | 11% | — |
| DTBench | 80% | — |
| LMCA | 40.8% | — |
Math GPT-5.4 mini leads
GPT-5.4 mini: 45.4 (#63), Kimi K2 (Jul 2025): 40.1 (#81)
| Benchmark | GPT-5.4 mini | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Math | 1418 | 1396 |
| FrontierMath (Feb 2025 set) | 28.3% | 21.4% |
| FrontierMath Tier 4 (v1) | 2.1% | 0% |
| FrontierMath (Tiers 1-3) | 51.2% | — |
| FrontierMath Tier 4 | 9.8% | — |
| OTIS Mock AIME 2024-2025 | 88.9% | — |
| ProofBench | 21% | — |
| Omni-MATH | — | 65.4% |
Knowledge GPT-5.4 mini leads
GPT-5.4 mini: 51.4 (#62), Kimi K2 (Jul 2025): 36.9 (#153)
| Benchmark | GPT-5.4 mini | Kimi K2 (Jul 2025) |
|---|---|---|
| Vectara Hallucination Rate | 5.5% | 17.9% |
| LMArena Expert | 1434 | 1363 |
| GPQA Diamond | 86.9% | — |
| SimpleQA Verified | 29.4% | — |
| MMLU-Pro | — | 81.9% |
| Confabulations | — | 20.4% |
| GPQA (HELM) | — | 65.3% |
Multimodal Not comparable
GPT-5.4 mini: 39.0 (#51), Kimi K2 (Jul 2025): —
| Benchmark | GPT-5.4 mini | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Vision | 1245 | — |
Multilingual GPT-5.4 mini leads
GPT-5.4 mini: 52.1 (#94), Kimi K2 (Jul 2025): 49.8 (#125)
| Benchmark | GPT-5.4 mini | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Non-English | 1405 | 1372 |
| LMArena Chinese | 1446 | 1416 |
| LMArena French | 1440 | 1379 |
| LMArena German | 1407 | 1386 |
| LMArena Japanese | 1375 | 1349 |
| LMArena Korean | 1369 | 1324 |
| LMArena Russian | 1417 | 1384 |
| LMArena Spanish | 1403 | 1383 |
Instruction Following GPT-5.4 mini leads
GPT-5.4 mini: 74.3 (#99), Kimi K2 (Jul 2025): 71.3 (#151)
| Benchmark | GPT-5.4 mini | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Instruction Following | 1405 | 1347 |
| IFEval | — | 85% |
Long Context GPT-5.4 mini leads
GPT-5.4 mini: 42.9 (#107), Kimi K2 (Jul 2025): 41.2 (#140)
| Benchmark | GPT-5.4 mini | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Longer Query | 1407 | 1353 |
| Fiction.LiveBench | — | 66.7% |
| CL-bench | — | 17.6% |
Writing & Preference GPT-5.4 mini leads
GPT-5.4 mini: 64.2 (#56), Kimi K2 (Jul 2025): 62.5 (#76)
| Benchmark | GPT-5.4 mini | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Text | 1411 | 1380 |
| LMArena Creative Writing | 1370 | 1349 |
| EQ-Bench Creative Writing | 1665 | 1666 |
| LMArena Multi-Turn | 1429 | 1371 |
| Short-Story Creative Writing | — | 85.6% |
| WildBench | — | 86.2% |
Frequently asked questions
Is GPT-5.4 mini better than Kimi K2 (Jul 2025)?
GPT-5.4 mini is the stronger model overall, scoring 44.8 to 40.9 on the Noometry Index. Kimi K2 (Jul 2025) costs 1.7× less per token, which makes it the better buy when GPT-5.4 mini's lead doesn't matter for your workload.
Which is cheaper, GPT-5.4 mini or Kimi K2 (Jul 2025)?
Kimi K2 (Jul 2025) is cheaper. It lists at $0.57 per million input tokens and $2.30 per million output tokens; GPT-5.4 mini lists at $0.75 and $4.50.
Is GPT-5.4 mini or Kimi K2 (Jul 2025) better for coding?
GPT-5.4 mini scores higher on coding benchmarks: 45.0 versus 43.1 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 mini does, with 400K tokens against 262K.
How many benchmarks do GPT-5.4 mini and Kimi K2 (Jul 2025) share?
26 benchmarks have published results for both models. GPT-5.4 mini has 46 scored results on Noometry and Kimi K2 (Jul 2025) has 42.