Model comparison
Kimi K2.5 (Reasoning) vs Soofi S 30B-A3B
Head-to-head evidence from 2 shared benchmark results across 1 category. Overall scores shown here use BenchLM's provisional ranking lane.
Evidence parity. Kimi K2.5 (Reasoning) and Soofi S 30B-A3B share 2 comparable benchmark results. 1 of 8 categories are comparable. 26 results are unique to Kimi K2.5 (Reasoning); 6 to Soofi S 30B-A3B.
Updated July 15, 2026- Shared results
- 2
- Kimi K2.5 (Reasoning) only
- 26
- Soofi S 30B-A3B only
- 6
- Comparable categories
- 1 / 8
Pick Kimi K2.5 (Reasoning) if you want the stronger benchmark profile. Soofi S 30B-A3B only becomes the better choice if you want the cheaper token bill or you need the larger 1M context window.
Confidence note. This is a partial-evidence comparison with 2 shared benchmark results across 1 evidence category; 1 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
Kimi K2.5 (Reasoning) is clearly ahead on the provisional aggregate, 70 to 45. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Kimi K2.5 (Reasoning)'s sharpest advantage is in knowledge, where it averages 87.2 against 49.9. The single biggest benchmark swing on the page is GPQA, 87.6% to 43.4%.
Kimi K2.5 (Reasoning) is also the more expensive model on tokens at $0.60 input / $3.00 output per 1M tokens, versus $0.00 input / $0.00 output per 1M tokens for Soofi S 30B-A3B. That is roughly Infinityx on output cost alone. Kimi K2.5 (Reasoning) is the reasoning model in the pair, while Soofi S 30B-A3B is not. That usually helps on harder chain-of-thought-heavy tests, but it can also mean more latency and more token spend in real use. Soofi S 30B-A3B gives you the larger context window at 1M, compared with 128K for Kimi K2.5 (Reasoning).
Category breakdown
Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.
| Category | Kimi K2.5 (Reasoning) | Δ | Soofi S 30B-A3B |
|---|---|---|---|
| Knowledge | Kimi K2.5 (Reasoning)87.2 | Margin← 37.3 | Soofi S 30B-A3B49.9 |
| Agentic | Kimi K2.5 (Reasoning)55.0 | MarginNo overlap | Soofi S 30B-A3BNot measured |
| Coding | Kimi K2.5 (Reasoning)76.8 | MarginNo overlap | Soofi S 30B-A3BNot measured |
| Multimodal | Kimi K2.5 (Reasoning)78.5 | MarginNo overlap | Soofi S 30B-A3BNot measured |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
GPQA
KnowledgeA 87.6%B 43.4%Winner: Kimi K2.5 (Reasoning)Δ 44.2GPQA: Kimi K2.5 (Reasoning) scored 87.6%; Soofi S 30B-A3B scored 43.4%. Kimi K2.5 (Reasoning) wins this benchmark. - Source ↗
MMLU-Pro
KnowledgeA 87.1%B 51.4%Winner: Kimi K2.5 (Reasoning)Δ 35.7MMLU-Pro: Kimi K2.5 (Reasoning) scored 87.1%; Soofi S 30B-A3B scored 51.4%. Kimi K2.5 (Reasoning) wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | Kimi K2.5 (Reasoning) | Soofi S 30B-A3B | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | Kimi K2.5 (Reasoning)$0.6 input / $3 output | Soofi S 30B-A3B$0 input / $0 output | Soofi S 30B-A3B has the lower combined listed price. |
| Generation speedtokens per second | Kimi K2.5 (Reasoning)Not available | Soofi S 30B-A3BNot available | A complete speed comparison is not available. |
| First-answer latencyseconds to first token | Kimi K2.5 (Reasoning)Not available | Soofi S 30B-A3BNot available | A complete latency comparison is not available. |
| Context windowmaximum listed tokens | Kimi K2.5 (Reasoning)128K | Soofi S 30B-A3B1M | Soofi S 30B-A3B lists the larger context window. |
Benchmark Deep Dive
Agentic8 benchmarks
| Benchmark | Kimi K2.5 (Reasoning) | Soofi S 30B-A3B | Result |
|---|---|---|---|
| Terminal-Bench 2.0Source | 50.8% | — | Not comparable |
| BrowseCompSource | 60.6% | — | Not comparable |
| APEX-Agents-AASource | 11.5% | — | Not comparable |
| τ²-bench resultsSource | 95.9% | — | Not comparable |
| Gert LabsSource | 32.58% | — | Not comparable |
| AA Agentic IndexSource | 21.7% | — | Not comparable |
| GDPval-AASource | 25.4% | — | Not comparable |
| GDPval-AASource | 1009 | — | Not comparable |
Coding6 benchmarks
| Benchmark | Kimi K2.5 (Reasoning) | Soofi S 30B-A3B | Result |
|---|---|---|---|
| SWE-bench VerifiedSource | 76.8% | — | Not comparable |
| Vibe Code BenchSource | 17.54% | — | Not comparable |
| Terminal-Bench HardSource | 34.8% | — | Not comparable |
| AA-SciCodeSource | 49.0% | — | Not comparable |
| AA Coding IndexSource | 46.8% | — | Not comparable |
| HumanEvalSource | — | 73.8% | Not comparable |
Reasoning4 benchmarks
KnowledgeKimi K2.5 (Reasoning) wins10 benchmarks
| Benchmark | Kimi K2.5 (Reasoning) | Soofi S 30B-A3B | Result |
|---|---|---|---|
| GPQASource | 87.6% | 43.4% | Kimi K2.5 (Reasoning) leads |
| MMLU-ProSource | 87.1% | 51.4% | Kimi K2.5 (Reasoning) leads |
| Artificial Analysis Intelligence IndexSource | 35.4% | — | Not comparable |
| AA-GPQA DiamondSource | 87.9% | — | Not comparable |
| AA-HLESource | 29.4% | — | Not comparable |
| AA-Omniscience IndexSource | -8.1% | — | Not comparable |
| AA-Omniscience AccuracySource | 34.3% | — | Not comparable |
| AA-Omniscience Hallucination RateSource | 64.6% | — | Not comparable |
| GPQA-DSource | — | 43.4% | Not comparable |
| AGIEvalSource | — | 66.9% | Not comparable |
Math2 benchmarks
Multimodal3 benchmarks
Inst. Following1 benchmarks
| Benchmark | Kimi K2.5 (Reasoning) | Soofi S 30B-A3B | Result |
|---|---|---|---|
| AA-IFBenchSource | 70.2% | — | Not comparable |
Frequently Asked Questions (2)
Which is better, Kimi K2.5 (Reasoning) or Soofi S 30B-A3B?
Kimi K2.5 (Reasoning) is ahead on BenchLM's provisional leaderboard, 70 to 45. The biggest single separator in this matchup is GPQA, where the scores are 87.6% and 43.4%.
Which is better for knowledge tasks, Kimi K2.5 (Reasoning) or Soofi S 30B-A3B?
Kimi K2.5 (Reasoning) has the edge for knowledge tasks in this comparison, averaging 87.2 versus 49.9. Inside this category, GPQA is the benchmark that creates the most daylight between them.
Related Comparisons
The AI models change fast. We track them for you.
A weekly brief for engineers and researchers covering new models, ranking shifts, and pricing changes.
Free. No spam. Unsubscribe anytime.