Model comparison
GPT-5.6 Terra vs Kimi K2.5
GPT-5.6 Terra is the stronger model overall, scoring 59.1 to 47.8 on the Noometry Index. Kimi K2.5 costs 5.0× less per token, which makes it the better buy when GPT-5.6 Terra's lead doesn't matter for your workload.
Last verified . 36 shared benchmarks.
Summary
- They share 36 benchmarks with published results for both. GPT-5.6 Terra scores higher in 9 categories and Kimi K2.5 in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Terra leads 81.5 to 50.3.
- The biggest single-benchmark swing is ARC-AGI-2: 83.9% for GPT-5.6 Terra and 11.8% for Kimi K2.5.
- Kimi K2.5 is cheaper at $0.45 / $2.25 per million input/output tokens, against $2 / $12 for GPT-5.6 Terra.
- GPT-5.6 Terra accepts more context: 1.05M tokens versus 262K.
- Kimi K2.5 has downloadable open weights; the other is API-only.
Side by side
| GPT-5.6 Terra | Kimi K2.5 | |
|---|---|---|
| Provider | OpenAI | Moonshot AI |
| Noometry Index | 59.1 | 47.8 |
| Released | 2026-07-09 | 2026-01-27 |
| Weights | Proprietary | Open |
| Context window | 1.05M | 262K |
| Max output | 128K | 262K |
| Input $ / M tokens | $2 | $0.45 |
| Output $ / M tokens | $12 | $2.25 |
| Results tracked | 52 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.6 Terra leads
GPT-5.6 Terra: 57.1 (#16), Kimi K2.5: 48.7 (#48)
| Benchmark | GPT-5.6 Terra | Kimi K2.5 |
|---|---|---|
| LMArena WebDev | 1520 | 1436 |
| SciCode | 55% | 49% |
| WeirdML | 78.3% | 45.6% |
| LMArena Coding | 1484 | 1474 |
| ALE-Bench | 1,951 | 821.65 |
| SWE-bench Verified | — | 73.8% |
| DeepSWE | 69.6% | — |
| FrontierCode | 41.3% | — |
| SWE-bench Verified (bash only) | — | 70.8% |
| CursorBench | 41.3% | — |
| SWE-bench Multilingual | — | 67.3% |
Agentic & Tool Use GPT-5.6 Terra leads
GPT-5.6 Terra: 40.2 (#23), Kimi K2.5: 34.0 (#44)
| Benchmark | GPT-5.6 Terra | Kimi K2.5 |
|---|---|---|
| Vending-Bench 2 | 7,343 | 1,198 |
| Terminal-Bench | — | 43.2% |
| APEX-Agents | 58.2% | — |
| OSWorld | — | 63.3% |
| BALROG | 53.2% | — |
| GDP.pdf | 24.7% | — |
Reasoning GPT-5.6 Terra leads
GPT-5.6 Terra: 60.6 (#19), Kimi K2.5: 31.2 (#74)
| Benchmark | GPT-5.6 Terra | Kimi K2.5 |
|---|---|---|
| ARC-AGI-2 | 83.9% | 11.8% |
| SimpleBench | 48.9% | 46.8% |
| Kagi LLM Benchmark | 51.3% | 78.5% |
| NYT Connections (extended) | 78.4% | 69.9% |
| ARC-AGI-1 | 96.5% | 65.3% |
| CritPt | 30% | 3.1% |
| Chess Puzzles | 54% | 12% |
| LMArena Hard Prompts | 1467 | 1454 |
| Epoch Capabilities Index | 159.31 | 148.05 |
| EnigmaEval | — | 3.4% |
| Thematic Generalization | — | 69.4% |
| Mystery Game Puzzles | 35% | — |
| DTBench | 93.3% | — |
| LMCA | 55% | — |
| Surface Evolver Bench | 83.8% | — |
Math GPT-5.6 Terra leads
GPT-5.6 Terra: 81.5 (#10), Kimi K2.5: 50.3 (#46)
| Benchmark | GPT-5.6 Terra | Kimi K2.5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 99.7% | 92.2% |
| LMArena Math | 1463 | 1470 |
| FrontierMath (Tiers 1-3) | 86% | — |
| FrontierMath Tier 4 | 70.7% | — |
| MathArena Final-Answer Competitions | — | 62.3% |
| ProofBench | 74% | — |
| FrontierMath (Feb 2025 set) | — | 27.9% |
| FrontierMath Tier 4 (v1) | — | 4.2% |
Knowledge GPT-5.6 Terra leads
GPT-5.6 Terra: 61.3 (#27), Kimi K2.5: 53.6 (#52)
| Benchmark | GPT-5.6 Terra | Kimi K2.5 |
|---|---|---|
| GPQA Diamond | 93.3% | 87.6% |
| SimpleQA Verified | 43.2% | 34.3% |
| LMArena Expert | 1492 | 1466 |
| Humanity's Last Exam | — | 24.4% |
| Vectara Hallucination Rate | — | 14.2% |
Multimodal GPT-5.6 Terra leads
GPT-5.6 Terra: 47.5 (#10), Kimi K2.5: 40.4 (#37)
| Benchmark | GPT-5.6 Terra | Kimi K2.5 |
|---|---|---|
| LMArena Vision | 1271 | 1269 |
| LMArena Document | 1472 | 1430 |
| Blueprint-Bench 2 | 30.8% | — |
| Furniture Assembly | 54.2% | — |
Multilingual Too close to call
GPT-5.6 Terra: 54.5 (#42), Kimi K2.5: 54.1 (#50)
| Benchmark | GPT-5.6 Terra | Kimi K2.5 |
|---|---|---|
| LMArena Non-English | 1439 | 1433 |
| LMArena Chinese | 1514 | 1495 |
| LMArena French | 1476 | 1454 |
| LMArena German | 1462 | 1441 |
| LMArena Japanese | 1446 | 1422 |
| LMArena Korean | 1420 | 1412 |
| LMArena Russian | 1450 | 1436 |
| LMArena Spanish | 1455 | 1450 |
Instruction Following GPT-5.6 Terra leads
GPT-5.6 Terra: 76.6 (#38), Kimi K2.5: 75.5 (#60)
| Benchmark | GPT-5.6 Terra | Kimi K2.5 |
|---|---|---|
| LMArena Instruction Following | 1453 | 1431 |
Long Context Kimi K2.5 leads
GPT-5.6 Terra: 44.4 (#64), Kimi K2.5: 52.1 (#7)
| Benchmark | GPT-5.6 Terra | Kimi K2.5 |
|---|---|---|
| LMArena Longer Query | 1452 | 1445 |
| Fiction.LiveBench | — | 86.1% |
| CL-bench | — | 19.3% |
| CL-bench Life | — | 13.2% |
Writing & Preference GPT-5.6 Terra leads
GPT-5.6 Terra: 70.4 (#21), Kimi K2.5: 65.3 (#51)
| Benchmark | GPT-5.6 Terra | Kimi K2.5 |
|---|---|---|
| LMArena Text | 1446 | 1445 |
| LMArena Creative Writing | 1411 | 1423 |
| EQ-Bench Creative Writing | 1855 | 1579 |
| LMArena Multi-Turn | 1451 | 1444 |
| EQ-Bench 4 | 1234 | — |
Frequently asked questions
Is GPT-5.6 Terra better than Kimi K2.5?
GPT-5.6 Terra is the stronger model overall, scoring 59.1 to 47.8 on the Noometry Index. Kimi K2.5 costs 5.0× less per token, which makes it the better buy when GPT-5.6 Terra's lead doesn't matter for your workload.
Which is cheaper, GPT-5.6 Terra 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-5.6 Terra lists at $2 and $12.
Is GPT-5.6 Terra or Kimi K2.5 better for coding?
GPT-5.6 Terra scores higher on coding benchmarks: 57.1 versus 48.7 in the Noometry coding category.
Which has the bigger context window?
GPT-5.6 Terra does, with 1.05M tokens against 262K.
How many benchmarks do GPT-5.6 Terra and Kimi K2.5 share?
36 benchmarks have published results for both models. GPT-5.6 Terra has 52 scored results on Noometry and Kimi K2.5 has 51.