Model comparison
Claude Haiku 4.5 vs Kimi K2 (Jul 2025)
Kimi K2 (Jul 2025) is the stronger model overall, scoring 41.2 to 39.5 on the Noometry Index.
Last verified . 32 shared benchmarks.
Summary
- They share 32 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 7 categories and Kimi K2 (Jul 2025) in 2 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Kimi K2 (Jul 2025) leads 23.3 to 15.1.
- The biggest single-benchmark swing is Berkeley Function Calling Leaderboard: 68.7% for Claude Haiku 4.5 and 59.1% for Kimi K2 (Jul 2025).
- Kimi K2 (Jul 2025) is cheaper at $0.57 / $2.30 per million input/output tokens, against $1 / $5 for Claude Haiku 4.5.
- Kimi K2 (Jul 2025) accepts more context: 262K tokens versus 200K.
- Kimi K2 (Jul 2025) has downloadable open weights; the other is API-only.
Side by side
| Claude Haiku 4.5 | Kimi K2 (Jul 2025) | |
|---|---|---|
| Provider | Anthropic | Moonshot AI |
| Noometry Index | 39.5 | 41.2 |
| Released | 2025-10-15 | 2025-07-12 |
| Weights | Proprietary | Open |
| Context window | 200K | 262K |
| Max output | 64K | 262K |
| Input $ / M tokens | $1 | $0.57 |
| Output $ / M tokens | $5 | $2.30 |
| Results tracked | 53 | 42 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Haiku 4.5 leads
Claude Haiku 4.5: 44.0 (#78), Kimi K2 (Jul 2025): 42.4 (#102)
| Benchmark | Claude Haiku 4.5 | Kimi K2 (Jul 2025) |
|---|---|---|
| SWE-bench Verified (bash only) | 66.6% | 63.4% |
| WeirdML | 45.4% | 42.8% |
| LMArena Coding | 1453 | 1399 |
| ALE-Bench | 653.48 | 597.5 |
| Aider Polyglot | — | 59.1% |
| LMArena WebDev | 1330 | — |
| SWE-bench Multilingual | 64.7% | — |
| SciCode | 43.3% | — |
| GSO | — | 4.9% |
Agentic & Tool Use Claude Haiku 4.5 leads
Claude Haiku 4.5: 33.6 (#52), Kimi K2 (Jul 2025): 32.4 (#64)
| Benchmark | Claude Haiku 4.5 | Kimi K2 (Jul 2025) |
|---|---|---|
| Terminal-Bench | 35.5% | 35.7% |
| Berkeley Function Calling Leaderboard | 68.7% | 59.1% |
| DeepResearch Bench | 45.5% | — |
| BALROG | 31.2% | — |
| ExploitBench | 13.7% | — |
| METR Time Horizons | — | 59.2% |
| Vending-Bench 2 | 458.89 | — |
Reasoning Kimi K2 (Jul 2025) leads
Claude Haiku 4.5: 15.1 (#320), Kimi K2 (Jul 2025): 23.3 (#179)
| Benchmark | Claude Haiku 4.5 | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Hard Prompts | 1420 | 1384 |
| Epoch Capabilities Index | 142.41 | 146.01 |
| ForecastBench | 61.4 | 60.2 |
| ARC-AGI-2 | 4% | — |
| SimpleBench | — | 26.3% |
| Kagi LLM Benchmark | — | 64.4% |
| NYT Connections (extended) | 14.3% | — |
| ARC-AGI-1 | 47.7% | — |
| CritPt | 0% | — |
| Chess Puzzles | 8% | — |
| DTBench | 73.6% | — |
| LMCA | 30.9% | — |
Math Claude Haiku 4.5 leads
Claude Haiku 4.5: 44.9 (#78), Kimi K2 (Jul 2025): 42.7 (#83)
| Benchmark | Claude Haiku 4.5 | Kimi K2 (Jul 2025) |
|---|---|---|
| Omni-MATH | 56.1% | 65.4% |
| LMArena Math | 1396 | 1397 |
| FrontierMath (Feb 2025 set) | 5.9% | 21.4% |
| FrontierMath Tier 4 (v1) | 2.1% | 0% |
| OTIS Mock AIME 2024-2025 | 66.7% | — |
| MATH Level 5 | 96.4% | — |
Knowledge Too close to call
Claude Haiku 4.5: 37.7 (#153), Kimi K2 (Jul 2025): 37.3 (#157)
| Benchmark | Claude Haiku 4.5 | Kimi K2 (Jul 2025) |
|---|---|---|
| MMLU-Pro | 77.7% | 81.9% |
| Vectara Hallucination Rate | 9.8% | 17.9% |
| GPQA (HELM) | 60.5% | 65.3% |
| LMArena Expert | 1442 | 1365 |
| GPQA Diamond | 71.2% | — |
| SimpleQA Verified | 13.2% | — |
| Confabulations | — | 20.4% |
Multimodal Not comparable
Claude Haiku 4.5: 26.8 (#118), Kimi K2 (Jul 2025): —
| Benchmark | Claude Haiku 4.5 | Kimi K2 (Jul 2025) |
|---|---|---|
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |
Multilingual Too close to call
Claude Haiku 4.5: 49.9 (#129), Kimi K2 (Jul 2025): 49.6 (#130)
| Benchmark | Claude Haiku 4.5 | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Non-English | 1377 | 1372 |
| LMArena Chinese | 1417 | 1415 |
| LMArena French | 1408 | 1379 |
| LMArena German | 1375 | 1387 |
| LMArena Japanese | 1339 | 1349 |
| LMArena Korean | 1347 | 1325 |
| LMArena Russian | 1381 | 1385 |
| LMArena Spanish | 1420 | 1386 |
Instruction Following Too close to call
Claude Haiku 4.5: 71.4 (#149), Kimi K2 (Jul 2025): 71.1 (#156)
| Benchmark | Claude Haiku 4.5 | Kimi K2 (Jul 2025) |
|---|---|---|
| IFEval | 80.1% | 85% |
| LMArena Instruction Following | 1414 | 1348 |
Long Context Claude Haiku 4.5 leads
Claude Haiku 4.5: 43.6 (#92), Kimi K2 (Jul 2025): 41.2 (#145)
| Benchmark | Claude Haiku 4.5 | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Longer Query | 1427 | 1353 |
| Fiction.LiveBench | — | 66.7% |
| CL-bench | — | 17.6% |
Writing & Preference Kimi K2 (Jul 2025) leads
Claude Haiku 4.5: 57.9 (#123), Kimi K2 (Jul 2025): 62.3 (#78)
| Benchmark | Claude Haiku 4.5 | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Text | 1396 | 1380 |
| LMArena Creative Writing | 1372 | 1350 |
| WildBench | 83.9% | 86.2% |
| LMArena Multi-Turn | 1409 | 1371 |
| Short-Story Creative Writing | — | 85.6% |
| EQ-Bench Creative Writing | — | 1666 |
| EQ-Bench 4 | 1064 | — |
Frequently asked questions
Is Claude Haiku 4.5 better than Kimi K2 (Jul 2025)?
Kimi K2 (Jul 2025) is the stronger model overall, scoring 41.2 to 39.5 on the Noometry Index.
Which is cheaper, Claude Haiku 4.5 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; Claude Haiku 4.5 lists at $1 and $5.
Is Claude Haiku 4.5 or Kimi K2 (Jul 2025) better for coding?
Claude Haiku 4.5 scores higher on coding benchmarks: 44.0 versus 42.4 in the Noometry coding category.
Which has the bigger context window?
Kimi K2 (Jul 2025) does, with 262K tokens against 200K.
How many benchmarks do Claude Haiku 4.5 and Kimi K2 (Jul 2025) share?
32 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and Kimi K2 (Jul 2025) has 42.