AI BENCHY Compare
DeepSeek: DeepSeek V3.2 vs MoonshotAI: Kimi K2.5
Last updated at: 2026-03-03
| Metric | DeepSeek: DeepSeek V3.2 medium Release: 2025-12-01 | MoonshotAI: Kimi K2.5 none Release: 2026-01-27 |
|---|---|---|
| Rank | #12 | #35 |
| Avg Score | 6.98 | 4.07 |
| Consistency | 8.75 | 8.92 |
| Cost per result | 0.193 | 0.232 |
| Total Cost | $0.018 | $0.010 |
| Tests Correct | 9/14 | 4/14 |
| Attempt pass rate | 71.4% | 35.7% |
| Flaky tests | 2 | 2 |
| Output Tokens | 6,753 | 1,915 |
| Reasoning Tokens | 30,427 | 0 |
Category Breakdown
| Anti-AI Tricks | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|
| DeepSeek: DeepSeek V3.2 | 7.00 | 9.86 | 66.7% | 0 | 1,171 | 4,893 | |
| MoonshotAI: Kimi K2.5 | 2.67 | 7.86 | 11.1% | 1 | 363 | 0 |
| Data parsing and extraction | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|
| DeepSeek: DeepSeek V3.2 | 9.88 | 10.00 | 100.0% | 0 | 207 | 7,693 | |
| MoonshotAI: Kimi K2.5 | 5.50 | 5.81 | 83.3% | 1 | 995 | 0 |
| Domain specific | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|
| DeepSeek: DeepSeek V3.2 | 4.00 | 7.21 | 44.4% | 1 | 3,081 | 7,856 | |
| MoonshotAI: Kimi K2.5 | 4.00 | 10.00 | 33.3% | 0 | 29 | 0 |
| Instructions following | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|
| DeepSeek: DeepSeek V3.2 | 7.00 | 9.84 | 50.0% | 0 | 1,397 | 2,845 | |
| MoonshotAI: Kimi K2.5 | 5.00 | 9.99 | 50.0% | 0 | 61 | 0 |
| Puzzle Solving | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|
| DeepSeek: DeepSeek V3.2 | 7.00 | 7.21 | 88.9% | 1 | 390 | 6,281 | |
| MoonshotAI: Kimi K2.5 | 2.00 | 9.92 | 0.0% | 0 | 247 | 0 |
| Tool Calling | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|
| DeepSeek: DeepSeek V3.2 | 10.00 | 10.00 | 100.0% | 0 | 507 | 859 | |
| MoonshotAI: Kimi K2.5 | 10.00 | 10.00 | 100.0% | 0 | 220 | 0 |
Quick Compare
Switch Comparison Pair
Kimi K2.5nonevsGLM 4.7 FlashmediumDeepSeek V3.2mediumvsGemini 3 Flash PreviewnoneKimi K2.5nonevsQwen3 Coder NextmediumClaude Sonnet 4.6nonevsDeepSeek V3.2mediumDeepSeek V3.2mediumvsQwen3.5 Plus 2026-02-15noneDeepSeek V3.2mediumvsGLM 5noneDeepSeek V3.2mediumvsGemini 3 Flash PreviewlowMiniMax M2.5mediumvsKimi K2.5noneKimi K2.5nonevsgpt-oss-120bmediumFree AvailableKimi K2.5nonevsQwen3.5-FlashmediumKimi K2.5nonevsGPT-5 NanomediumDeepSeek V3.2mediumvsQwen3.5-122B-A10Bnone