Model comparison
GPT-5.4 Pro vs Kimi K2.5
GPT-5.4 Pro is the stronger model overall, scoring 58.8 to 47.8 on the Noometry Index. Kimi K2.5 costs 75× less per token, which makes it the better buy when GPT-5.4 Pro's lead doesn't matter for your workload.
Last verified . 14 shared benchmarks.
Summary
- They share 14 benchmarks with published results for both. GPT-5.4 Pro scores higher in 3 categories and Kimi K2.5 in 1 category; 4 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.4 Pro leads 71.0 to 31.2.
- The biggest single-benchmark swing is ARC-AGI-2: 83.3% for GPT-5.4 Pro and 11.8% for Kimi K2.5.
- Kimi K2.5 is cheaper at $0.45 / $2.25 per million input/output tokens, against $30 / $180 for GPT-5.4 Pro.
- GPT-5.4 Pro accepts more context: 1.05M tokens versus 262K.
- Kimi K2.5 has downloadable open weights; the other is API-only.
Side by side
| GPT-5.4 Pro | Kimi K2.5 | |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 58.8 | 47.8 |
| Released | 2026-03-05 | 2026-01-27 |
| Weights | Proprietary | Open |
| Context window | 1.05M | 262K |
| Max output | 128K | 262K |
| Input $ / M tokens | $30 | $0.45 |
| Output $ / M tokens | $180 | $2.25 |
| Results tracked | 16 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.5 leads
GPT-5.4 Pro: 42.6 (#89), Kimi K2.5: 48.7 (#48)
| Benchmark | GPT-5.4 Pro | Kimi K2.5 |
|---|---|---|
| WeirdML | 57.4% | 45.6% |
| SWE-bench Verified | — | 73.8% |
| SWE-bench Verified (bash only) | — | 70.8% |
| LMArena WebDev | — | 1436 |
| SWE-bench Multilingual | — | 67.3% |
| SciCode | — | 49% |
| LMArena Coding | — | 1474 |
| ALE-Bench | — | 821.65 |
Agentic & Tool Use Not comparable
GPT-5.4 Pro: —, Kimi K2.5: 34.0 (#44)
| Benchmark | GPT-5.4 Pro | Kimi K2.5 |
|---|---|---|
| Terminal-Bench | — | 43.2% |
| OSWorld | — | 63.3% |
| Vending-Bench 2 | — | 1,198 |
Reasoning GPT-5.4 Pro leads
GPT-5.4 Pro: 71.0 (#10), Kimi K2.5: 31.2 (#74)
| Benchmark | GPT-5.4 Pro | Kimi K2.5 |
|---|---|---|
| ARC-AGI-2 | 83.3% | 11.8% |
| SimpleBench | 74.1% | 46.8% |
| ARC-AGI-1 | 94.5% | 65.3% |
| CritPt | 30% | 3.1% |
| Chess Puzzles | 58.6% | 12% |
| EnigmaEval | 23.8% | 3.4% |
| Epoch Capabilities Index | 159.12 | 148.05 |
| Kagi LLM Benchmark | — | 78.5% |
| NYT Connections (extended) | — | 69.9% |
| Thematic Generalization | — | 69.4% |
| LMArena Hard Prompts | — | 1454 |
Math GPT-5.4 Pro leads
GPT-5.4 Pro: 72.0 (#17), Kimi K2.5: 50.3 (#46)
| Benchmark | GPT-5.4 Pro | Kimi K2.5 |
|---|---|---|
| FrontierMath (Feb 2025 set) | 50% | 27.9% |
| FrontierMath Tier 4 (v1) | 37.5% | 4.2% |
| FrontierMath (Tiers 1-3) | 82.5% | — |
| FrontierMath Tier 4 | 58.5% | — |
| MathArena Final-Answer Competitions | — | 62.3% |
| OTIS Mock AIME 2024-2025 | — | 92.2% |
| LMArena Math | — | 1470 |
Knowledge GPT-5.4 Pro leads
GPT-5.4 Pro: 68.3 (#6), Kimi K2.5: 53.6 (#52)
| Benchmark | GPT-5.4 Pro | Kimi K2.5 |
|---|---|---|
| GPQA Diamond | 94.6% | 87.6% |
| Humanity's Last Exam | 44.3% | 24.4% |
| SimpleQA Verified | 46.3% | 34.3% |
| Vectara Hallucination Rate | 8.3% | 14.2% |
| LMArena Expert | — | 1466 |
Multimodal Not comparable
GPT-5.4 Pro: —, Kimi K2.5: 40.4 (#37)
| Benchmark | GPT-5.4 Pro | Kimi K2.5 |
|---|---|---|
| LMArena Vision | — | 1269 |
| LMArena Document | — | 1430 |
Multilingual Not comparable
GPT-5.4 Pro: —, Kimi K2.5: 54.1 (#50)
| Benchmark | GPT-5.4 Pro | Kimi K2.5 |
|---|---|---|
| LMArena Non-English | — | 1433 |
| LMArena Chinese | — | 1495 |
| LMArena French | — | 1454 |
| LMArena German | — | 1441 |
| LMArena Japanese | — | 1422 |
| LMArena Korean | — | 1412 |
| LMArena Russian | — | 1436 |
| LMArena Spanish | — | 1450 |
Instruction Following Not comparable
GPT-5.4 Pro: —, Kimi K2.5: 75.5 (#60)
| Benchmark | GPT-5.4 Pro | Kimi K2.5 |
|---|---|---|
| LMArena Instruction Following | — | 1431 |
Long Context Not comparable
GPT-5.4 Pro: —, Kimi K2.5: 52.1 (#7)
| Benchmark | GPT-5.4 Pro | Kimi K2.5 |
|---|---|---|
| Fiction.LiveBench | — | 86.1% |
| CL-bench | — | 19.3% |
| CL-bench Life | — | 13.2% |
| LMArena Longer Query | — | 1445 |
Writing & Preference Not comparable
GPT-5.4 Pro: —, Kimi K2.5: 65.3 (#51)
| Benchmark | GPT-5.4 Pro | Kimi K2.5 |
|---|---|---|
| LMArena Text | — | 1445 |
| LMArena Creative Writing | — | 1423 |
| EQ-Bench Creative Writing | — | 1579 |
| LMArena Multi-Turn | — | 1444 |
Frequently asked questions
Is GPT-5.4 Pro better than Kimi K2.5?
GPT-5.4 Pro is the stronger model overall, scoring 58.8 to 47.8 on the Noometry Index. Kimi K2.5 costs 75× less per token, which makes it the better buy when GPT-5.4 Pro's lead doesn't matter for your workload.
Which is cheaper, GPT-5.4 Pro or Kimi K2.5?
Kimi K2.5 is cheaper. It lists at $0.45 per million input tokens and $2.25 per million output tokens; GPT-5.4 Pro lists at $30 and $180.
Is GPT-5.4 Pro or Kimi K2.5 better for coding?
Kimi K2.5 scores higher on coding benchmarks: 48.7 versus 42.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 Pro does, with 1.05M tokens against 262K.
How many benchmarks do GPT-5.4 Pro and Kimi K2.5 share?
14 benchmarks have published results for both models. GPT-5.4 Pro has 16 scored results on Noometry and Kimi K2.5 has 51.