Model comparison
Claude 3 Haiku vs Kimi K2.5
Kimi K2.5 is the stronger model overall, scoring 47.8 to 25.7 on the Noometry Index.
Last verified . 24 shared benchmarks.
Summary
- They share 24 benchmarks with published results for both. Claude 3 Haiku scores higher in 0 categories and Kimi K2.5 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K2.5 leads 50.3 to 9.2.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 1.8% for Claude 3 Haiku and 92.2% for Kimi K2.5.
- Kimi K2.5 has downloadable open weights; the other is API-only.
Side by side
| Claude 3 Haiku | Kimi K2.5 | |
|---|---|---|
| Provider | Anthropic | Moonshot AI |
| Noometry Index | 25.7 | 47.8 |
| Released | 2024-03-07 | 2026-01-27 |
| Weights | Proprietary | Open |
| Context window | — | 262K |
| Max output | — | 262K |
| Input $ / M tokens | — | $0.45 |
| Output $ / M tokens | — | $2.25 |
| Results tracked | 37 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.5 leads
Claude 3 Haiku: 26.1 (#320), Kimi K2.5: 48.7 (#48)
| Benchmark | Claude 3 Haiku | Kimi K2.5 |
|---|---|---|
| WeirdML | 9.8% | 45.6% |
| LMArena Coding | 1199 | 1474 |
| SWE-bench Verified | — | 73.8% |
| SWE-bench Verified (bash only) | — | 70.8% |
| LMArena WebDev | — | 1436 |
| SWE-bench Multilingual | — | 67.3% |
| SciCode | — | 49% |
| BigCodeBench Instruct | 39.4% | — |
| BigCodeBench Complete | 50.1% | — |
| CadEval | 12% | — |
| ALE-Bench | — | 821.65 |
| HumanEval+ | 68.9% | — |
| MBPP+ | 68.8% | — |
Agentic & Tool Use Not comparable
Claude 3 Haiku: —, Kimi K2.5: 34.0 (#44)
| Benchmark | Claude 3 Haiku | Kimi K2.5 |
|---|---|---|
| Terminal-Bench | — | 43.2% |
| OSWorld | — | 63.3% |
| Vending-Bench 2 | — | 1,198 |
Reasoning Kimi K2.5 leads
Claude 3 Haiku: 15.7 (#303), Kimi K2.5: 31.2 (#74)
| Benchmark | Claude 3 Haiku | Kimi K2.5 |
|---|---|---|
| Kagi LLM Benchmark | 34.2% | 78.5% |
| LMArena Hard Prompts | 1174 | 1454 |
| Epoch Capabilities Index | 118.3 | 148.05 |
| ARC-AGI-2 | — | 11.8% |
| SimpleBench | — | 46.8% |
| NYT Connections (extended) | — | 69.9% |
| ARC-AGI-1 | — | 65.3% |
| CritPt | — | 3.1% |
| Chess Puzzles | — | 12% |
| EnigmaEval | — | 3.4% |
| Thematic Generalization | — | 69.4% |
| DTBench | 50.1% | — |
| LMCA | 8.8% | — |
| ForecastBench | 53.2 | — |
| WinoGrande | 74.2% | — |
Math Kimi K2.5 leads
Claude 3 Haiku: 9.2 (#313), Kimi K2.5: 50.3 (#46)
| Benchmark | Claude 3 Haiku | Kimi K2.5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.8% | 92.2% |
| LMArena Math | 1188 | 1470 |
| MathArena Final-Answer Competitions | — | 62.3% |
| MATH Level 5 | 14.9% | — |
| FrontierMath (Feb 2025 set) | — | 27.9% |
| FrontierMath Tier 4 (v1) | — | 4.2% |
Knowledge Kimi K2.5 leads
Claude 3 Haiku: 17.3 (#280), Kimi K2.5: 53.6 (#52)
| Benchmark | Claude 3 Haiku | Kimi K2.5 |
|---|---|---|
| GPQA Diamond | 36.3% | 87.6% |
| LMArena Expert | 1148 | 1466 |
| Humanity's Last Exam | — | 24.4% |
| SimpleQA Verified | — | 34.3% |
| Confabulations | 34.2% | — |
| Vectara Hallucination Rate | — | 14.2% |
| MMLU | 73.8% | — |
Multimodal Kimi K2.5 leads
Claude 3 Haiku: 23.1 (#123), Kimi K2.5: 40.4 (#37)
| Benchmark | Claude 3 Haiku | Kimi K2.5 |
|---|---|---|
| LMArena Vision | 950 | 1269 |
| LMArena Document | — | 1430 |
| ScienceQA | 72% | — |
Multilingual Kimi K2.5 leads
Claude 3 Haiku: 36.1 (#238), Kimi K2.5: 54.1 (#50)
| Benchmark | Claude 3 Haiku | Kimi K2.5 |
|---|---|---|
| LMArena Non-English | 1178 | 1433 |
| LMArena Chinese | 1155 | 1495 |
| LMArena French | 1194 | 1454 |
| LMArena German | 1174 | 1441 |
| LMArena Japanese | 1102 | 1422 |
| LMArena Korean | 1108 | 1412 |
| LMArena Russian | 1204 | 1436 |
| LMArena Spanish | 1165 | 1450 |
Instruction Following Kimi K2.5 leads
Claude 3 Haiku: 61.5 (#242), Kimi K2.5: 75.5 (#60)
| Benchmark | Claude 3 Haiku | Kimi K2.5 |
|---|---|---|
| LMArena Instruction Following | 1174 | 1431 |
Long Context Kimi K2.5 leads
Claude 3 Haiku: 36.1 (#232), Kimi K2.5: 52.1 (#7)
| Benchmark | Claude 3 Haiku | Kimi K2.5 |
|---|---|---|
| LMArena Longer Query | 1191 | 1445 |
| Fiction.LiveBench | — | 86.1% |
| CL-bench | — | 19.3% |
| CL-bench Life | — | 13.2% |
Writing & Preference Kimi K2.5 leads
Claude 3 Haiku: 29.8 (#286), Kimi K2.5: 65.3 (#51)
| Benchmark | Claude 3 Haiku | Kimi K2.5 |
|---|---|---|
| LMArena Text | 1195 | 1445 |
| LMArena Creative Writing | 1157 | 1423 |
| EQ-Bench Creative Writing | 717 | 1579 |
| LMArena Multi-Turn | 1190 | 1444 |
Frequently asked questions
Is Claude 3 Haiku better than Kimi K2.5?
Kimi K2.5 is the stronger model overall, scoring 47.8 to 25.7 on the Noometry Index.
Is Claude 3 Haiku or Kimi K2.5 better for coding?
Kimi K2.5 scores higher on coding benchmarks: 48.7 versus 26.1 in the Noometry coding category.
How many benchmarks do Claude 3 Haiku and Kimi K2.5 share?
24 benchmarks have published results for both models. Claude 3 Haiku has 37 scored results on Noometry and Kimi K2.5 has 51.