Model comparison
DeepSeek V4 Pro vs Kimi K2 (Jul 2025)
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 41.2 on the Noometry Index.
Last verified . 24 shared benchmarks.
DeepSeekSummary
- They share 24 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 9 categories and Kimi K2 (Jul 2025) in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Pro leads 56.5 to 23.3.
- The biggest single-benchmark swing is WeirdML: 66.2% for DeepSeek V4 Pro and 42.8% for Kimi K2 (Jul 2025).
- Both cost about the same: $0.66 input and $1.98 output per million tokens.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 262K.
Side by side
| DeepSeek V4 Pro | Kimi K2 (Jul 2025) | |
|---|---|---|
| Provider | DeepSeek | Moonshot AI |
| Noometry Index | 54.3 | 41.2 |
| Released | 2026-04-24 | 2025-07-12 |
| Weights | Open | Open |
| Context window | 1M | 262K |
| Max output | 393K | 262K |
| Input $ / M tokens | $0.66 | $0.57 |
| Output $ / M tokens | $1.98 | $2.30 |
| Results tracked | 48 | 42 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4 Pro leads
DeepSeek V4 Pro: 52.4 (#34), Kimi K2 (Jul 2025): 42.4 (#102)
| Benchmark | DeepSeek V4 Pro | Kimi K2 (Jul 2025) |
|---|---|---|
| WeirdML | 66.2% | 42.8% |
| LMArena Coding | 1470 | 1399 |
| ALE-Bench | 1,403 | 597.5 |
| SWE-bench Verified | 77.6% | — |
| FrontierCode | 28.6% | — |
| SWE-bench Verified (bash only) | — | 63.4% |
| Aider Polyglot | — | 59.1% |
| LMArena WebDev | 1582 | — |
| SciCode | 51% | — |
| GSO | — | 4.9% |
Agentic & Tool Use Too close to call
DeepSeek V4 Pro: 32.8 (#58), Kimi K2 (Jul 2025): 32.4 (#64)
| Benchmark | DeepSeek V4 Pro | Kimi K2 (Jul 2025) |
|---|---|---|
| Terminal-Bench | — | 35.7% |
| APEX-Agents | 47.3% | — |
| Berkeley Function Calling Leaderboard | — | 59.1% |
| METR Time Horizons | — | 59.2% |
| Vending-Bench 2 | 3,285 | — |
Reasoning DeepSeek V4 Pro leads
DeepSeek V4 Pro: 56.5 (#24), Kimi K2 (Jul 2025): 23.3 (#179)
| Benchmark | DeepSeek V4 Pro | Kimi K2 (Jul 2025) |
|---|---|---|
| Kagi LLM Benchmark | 53.5% | 64.4% |
| LMArena Hard Prompts | 1461 | 1384 |
| Epoch Capabilities Index | 155.31 | 146.01 |
| ForecastBench | 56.1 | 60.2 |
| ARC-AGI-2 | 61.3% | — |
| SimpleBench | — | 26.3% |
| NYT Connections (extended) | 91.3% | — |
| ARC-AGI-1 | 90.5% | — |
| CritPt | 18% | — |
| Chess Puzzles | 47% | — |
| Mystery Game Puzzles | 43% | — |
| DTBench | 93.9% | — |
| LMCA | 45.5% | — |
| Surface Evolver Bench | 40% | — |
Math DeepSeek V4 Pro leads
DeepSeek V4 Pro: 64.8 (#30), Kimi K2 (Jul 2025): 42.7 (#83)
| Benchmark | DeepSeek V4 Pro | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Math | 1455 | 1397 |
| FrontierMath (Tiers 1-3) | 64.6% | — |
| FrontierMath Tier 4 | 26.8% | — |
| MathArena Final-Answer Competitions | 76.6% | — |
| OTIS Mock AIME 2024-2025 | 98.6% | — |
| ProofBench | 50% | — |
| Omni-MATH | — | 65.4% |
| FrontierMath (Feb 2025 set) | — | 21.4% |
| FrontierMath Tier 4 (v1) | — | 0% |
Knowledge DeepSeek V4 Pro leads
DeepSeek V4 Pro: 59.5 (#31), Kimi K2 (Jul 2025): 37.3 (#157)
| Benchmark | DeepSeek V4 Pro | Kimi K2 (Jul 2025) |
|---|---|---|
| Vectara Hallucination Rate | 8.6% | 17.9% |
| LMArena Expert | 1464 | 1365 |
| GPQA Diamond | 91.7% | — |
| SimpleQA Verified | 52.9% | — |
| MMLU-Pro | — | 81.9% |
| Confabulations | — | 20.4% |
| GPQA (HELM) | — | 65.3% |
Multilingual DeepSeek V4 Pro leads
DeepSeek V4 Pro: 54.4 (#45), Kimi K2 (Jul 2025): 49.6 (#130)
| Benchmark | DeepSeek V4 Pro | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Non-English | 1439 | 1372 |
| LMArena Chinese | 1486 | 1415 |
| LMArena French | 1472 | 1379 |
| LMArena German | 1458 | 1387 |
| LMArena Japanese | 1445 | 1349 |
| LMArena Korean | 1447 | 1325 |
| LMArena Russian | 1453 | 1385 |
| LMArena Spanish | 1458 | 1386 |
Instruction Following DeepSeek V4 Pro leads
DeepSeek V4 Pro: 76.1 (#47), Kimi K2 (Jul 2025): 71.1 (#156)
| Benchmark | DeepSeek V4 Pro | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Instruction Following | 1448 | 1348 |
| IFEval | — | 85% |
Long Context DeepSeek V4 Pro leads
DeepSeek V4 Pro: 45.0 (#51), Kimi K2 (Jul 2025): 41.2 (#145)
| Benchmark | DeepSeek V4 Pro | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Longer Query | 1458 | 1353 |
| Fiction.LiveBench | — | 66.7% |
| CL-bench | — | 17.6% |
| CL-bench Life | 13.5% | — |
Writing & Preference DeepSeek V4 Pro leads
DeepSeek V4 Pro: 65.5 (#46), Kimi K2 (Jul 2025): 62.3 (#78)
| Benchmark | DeepSeek V4 Pro | Kimi K2 (Jul 2025) |
|---|---|---|
| LMArena Text | 1451 | 1380 |
| LMArena Creative Writing | 1446 | 1350 |
| EQ-Bench Creative Writing | 1553 | 1666 |
| LMArena Multi-Turn | 1467 | 1371 |
| Short-Story Creative Writing | — | 85.6% |
| WildBench | — | 86.2% |
| EQ-Bench 4 | 1166 | — |
Frequently asked questions
Is DeepSeek V4 Pro better than Kimi K2 (Jul 2025)?
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 41.2 on the Noometry Index.
Which is cheaper, DeepSeek V4 Pro or Kimi K2 (Jul 2025)?
DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; Kimi K2 (Jul 2025) lists at $0.57 and $2.30.
Is DeepSeek V4 Pro or Kimi K2 (Jul 2025) better for coding?
DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 42.4 in the Noometry coding category.
Which has the bigger context window?
DeepSeek V4 Pro does, with 1M tokens against 262K.
How many benchmarks do DeepSeek V4 Pro and Kimi K2 (Jul 2025) share?
24 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Kimi K2 (Jul 2025) has 42.