Model comparison
GPT-6 Luna vs Kimi K2.5
GPT-6 Luna is the stronger model overall, scoring 53.3 to 48.1 on the Noometry Index.
Last verified . 30 shared benchmarks.
Summary
- They share 30 benchmarks with published results for both. GPT-6 Luna scores higher in 5 categories and Kimi K2.5 in 5 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-6 Luna leads 76.1 to 51.8.
- The biggest single-benchmark swing is ARC-AGI-2: 59.3% for GPT-6 Luna and 11.8% for Kimi K2.5.
- GPT-6 Luna is cheaper at $0.10 / $0.50 per million input/output tokens, against $0.45 / $2.25 for Kimi K2.5.
- GPT-6 Luna accepts more context: 1.05M tokens versus 262K.
- Kimi K2.5 has downloadable open weights; the other is API-only.
Side by side
| GPT-6 Luna | Kimi K2.5 | |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 53.3 | 48.1 |
| Released | 2026-09-22 | 2026-01-27 |
| Weights | Proprietary | Open |
| Context window | 1.05M | 262K |
| Max output | 128K | 262K |
| Input $ / M tokens | $0.10 | $0.45 |
| Output $ / M tokens | $0.50 | $2.25 |
| Results tracked | 42 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Luna leads
GPT-6 Luna: 55.5 (#25), Kimi K2.5: 48.8 (#53)
| Benchmark | GPT-6 Luna | Kimi K2.5 |
|---|---|---|
| LMArena WebDev | 1581 | 1437 |
| SciCode | 54.6% | 49% |
| LMArena Coding | 1439 | 1474 |
| ALE-Bench | 1,577 | 821.65 |
| SWE-bench Verified | — | 73.8% |
| DeepSWE | 66.6% | — |
| FrontierCode | 42.4% | — |
| SWE-bench Verified (bash only) | — | 70.8% |
| SWE-bench Multilingual | — | 67.3% |
| WeirdML | — | 45.6% |
Agentic & Tool Use Too close to call
GPT-6 Luna: 33.3 (#54), Kimi K2.5: 34.2 (#48)
| Benchmark | GPT-6 Luna | Kimi K2.5 |
|---|---|---|
| Terminal-Bench | — | 43.2% |
| APEX-Agents | 44.3% | — |
| OSWorld | — | 63.3% |
| GDP.pdf | 23% | — |
| Vending-Bench 2 | — | 1,198 |
Reasoning GPT-6 Luna leads
GPT-6 Luna: 48.2 (#41), Kimi K2.5: 31.2 (#80)
| Benchmark | GPT-6 Luna | Kimi K2.5 |
|---|---|---|
| ARC-AGI-2 | 59.3% | 11.8% |
| NYT Connections (extended) | 68.7% | 69.9% |
| ARC-AGI-1 | 86.7% | 65.3% |
| CritPt | 19.4% | 3.1% |
| Chess Puzzles | 31% | 12% |
| LMArena Hard Prompts | 1411 | 1453 |
| Epoch Capabilities Index | 156.28 | 148.03 |
| SimpleBench | — | 46.8% |
| Kagi LLM Benchmark | — | 78.5% |
| EnigmaEval | — | 3.4% |
| Thematic Generalization | — | 69.4% |
| Mystery Game Puzzles | 7% | — |
| DTBench | 90.1% | — |
| LMCA | 44.5% | — |
Math GPT-6 Luna leads
GPT-6 Luna: 76.1 (#15), Kimi K2.5: 51.8 (#53)
| Benchmark | GPT-6 Luna | Kimi K2.5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 98.9% | 92.2% |
| LMArena Math | 1416 | 1470 |
| FrontierMath (Tiers 1-3) | 78.9% | — |
| FrontierMath Tier 4 | 56.1% | — |
| MathArena Final-Answer Competitions | — | 62.3% |
| ProofBench | 64% | — |
| FrontierMath (Feb 2025 set) | — | 27.9% |
| FrontierMath Tier 4 (v1) | — | 4.2% |
Knowledge GPT-6 Luna leads
GPT-6 Luna: 57.0 (#41), Kimi K2.5: 53.6 (#56)
| Benchmark | GPT-6 Luna | Kimi K2.5 |
|---|---|---|
| GPQA Diamond | 90.5% | 87.6% |
| SimpleQA Verified | 41.4% | 34.3% |
| LMArena Expert | 1444 | 1466 |
| Humanity's Last Exam | — | 24.4% |
| Vectara Hallucination Rate | — | 14.2% |
Multimodal GPT-6 Luna leads
GPT-6 Luna: 42.4 (#30), Kimi K2.5: 41.1 (#39)
| Benchmark | GPT-6 Luna | Kimi K2.5 |
|---|---|---|
| LMArena Vision | 1217 | 1269 |
| Blueprint-Bench 2 | 31.2% | — |
| Furniture Assembly | 44.2% | — |
| LMArena Document | — | 1430 |
Multilingual Kimi K2.5 leads
GPT-6 Luna: 50.5 (#117), Kimi K2.5: 53.9 (#53)
| Benchmark | GPT-6 Luna | Kimi K2.5 |
|---|---|---|
| LMArena Non-English | 1386 | 1433 |
| LMArena Chinese | 1433 | 1495 |
| LMArena French | 1420 | 1454 |
| LMArena German | 1369 | 1441 |
| LMArena Japanese | 1369 | 1421 |
| LMArena Korean | 1360 | 1410 |
| LMArena Russian | 1394 | 1435 |
| LMArena Spanish | 1393 | 1450 |
Instruction Following Kimi K2.5 leads
GPT-6 Luna: 74.3 (#99), Kimi K2.5: 75.3 (#64)
| Benchmark | GPT-6 Luna | Kimi K2.5 |
|---|---|---|
| LMArena Instruction Following | 1409 | 1431 |
Long Context Kimi K2.5 leads
GPT-6 Luna: 43.0 (#111), Kimi K2.5: 52.1 (#7)
| Benchmark | GPT-6 Luna | Kimi K2.5 |
|---|---|---|
| LMArena Longer Query | 1409 | 1445 |
| Fiction.LiveBench | — | 86.1% |
| CL-bench | — | 19.3% |
| CL-bench Life | — | 13.2% |
Writing & Preference Kimi K2.5 leads
GPT-6 Luna: 58.3 (#119), Kimi K2.5: 65.1 (#53)
| Benchmark | GPT-6 Luna | Kimi K2.5 |
|---|---|---|
| LMArena Text | 1391 | 1445 |
| LMArena Creative Writing | 1363 | 1423 |
| LMArena Multi-Turn | 1396 | 1444 |
| EQ-Bench Creative Writing | — | 1579 |
Frequently asked questions
Is GPT-6 Luna better than Kimi K2.5?
GPT-6 Luna is the stronger model overall, scoring 53.3 to 48.1 on the Noometry Index.
Which is cheaper, GPT-6 Luna or Kimi K2.5?
GPT-6 Luna is cheaper. It lists at $0.10 per million input tokens and $0.50 per million output tokens; Kimi K2.5 lists at $0.45 and $2.25.
Is GPT-6 Luna or Kimi K2.5 better for coding?
GPT-6 Luna scores higher on coding benchmarks: 55.5 versus 48.8 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Luna does, with 1.05M tokens against 262K.
How many benchmarks do GPT-6 Luna and Kimi K2.5 share?
30 benchmarks have published results for both models. GPT-6 Luna has 42 scored results on Noometry and Kimi K2.5 has 51.