Model comparison
Kimi K2.6 vs Qwen3.5 397B
Head-to-head evidence from 38 shared benchmark results across 7 categories. Overall scores shown here use BenchLM's provisional ranking lane.
Verified leaderboard positions: Kimi K2.6 #13; Qwen3.5 397B #23
Evidence parity. Kimi K2.6 and Qwen3.5 397B share 38 comparable benchmark results. 5 of 8 categories are comparable. 22 results are unique to Kimi K2.6; 18 to Qwen3.5 397B.
Updated July 13, 2026- Shared results
- 38
- Kimi K2.6 only
- 22
- Qwen3.5 397B only
- 18
- Comparable categories
- 5 / 8
Pick Kimi K2.6 if you want the stronger benchmark profile. Qwen3.5 397B only becomes the better choice if mathematics is the priority or you want the cheaper token bill.
Confidence note. This is a partial-evidence comparison with 38 shared benchmark results across 7 evidence categories; 5 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
Kimi K2.6 is clearly ahead on the provisional aggregate, 74 to 60. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Kimi K2.6's sharpest advantage is in agentic, where it averages 73.5 against 56.5. The single biggest benchmark swing on the page is BrowseComp, 83.2% to 62%. Qwen3.5 397B does hit back in mathematics, so the answer changes if that is the part of the workload you care about most.
Kimi K2.6 is also the more expensive model on tokens at $0.95 input / $4.00 output per 1M tokens, versus $0.60 input / $3.60 output per 1M tokens for Qwen3.5 397B. Kimi K2.6 is the reasoning model in the pair, while Qwen3.5 397B 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. Kimi K2.6 gives you the larger context window at 256K, compared with 128K for Qwen3.5 397B.
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 K2.6 | Δ | Qwen3.5 397B |
|---|---|---|---|
| Math | Kimi K2.667.1 | Margin→ 23.5 | Qwen3.5 397B90.6 |
| Agentic | Kimi K2.673.5 | Margin← 17.0 | Qwen3.5 397B56.5 |
| Knowledge | Kimi K2.642.2 | Margin→ 14.7 | Qwen3.5 397B56.9 |
| Coding | Kimi K2.672.6 | Margin← 12.3 | Qwen3.5 397B60.3 |
| Multimodal | Kimi K2.679.8 | Margin← 0.2 | Qwen3.5 397B79.6 |
| Reasoning | Kimi K2.6Not measured | MarginNo overlap | Qwen3.5 397B63.2 |
| Multilingual | Kimi K2.6Not measured | MarginNo overlap | Qwen3.5 397B84.7 |
| Inst. Following | Kimi K2.6Not measured | MarginNo overlap | Qwen3.5 397B92.6 |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
BrowseComp
AgenticA 83.2%B 62%Winner: Kimi K2.6Δ 21.2BrowseComp: Kimi K2.6 scored 83.2%; Qwen3.5 397B scored 62%. Kimi K2.6 wins this benchmark. - Source ↗
Terminal-Bench 2.0
AgenticA 66.7%B 52.5%Winner: Kimi K2.6Δ 14.2Terminal-Bench 2.0: Kimi K2.6 scored 66.7%; Qwen3.5 397B scored 52.5%. Kimi K2.6 wins this benchmark. - Source ↗
SWE-bench Pro
CodingA 58.6%B 50.9%Winner: Kimi K2.6Δ 7.7SWE-bench Pro: Kimi K2.6 scored 58.6%; Qwen3.5 397B scored 50.9%. Kimi K2.6 wins this benchmark. - Source ↗
HLE
KnowledgeA 34.7%B 28.7%Winner: Kimi K2.6Δ 6HLE: Kimi K2.6 scored 34.7%; Qwen3.5 397B scored 28.7%. Kimi K2.6 wins this benchmark. - Source ↗
HMMT Feb 2026
MathA 92.7%B 87.9%Winner: Kimi K2.6Δ 4.8HMMT Feb 2026: Kimi K2.6 scored 92.7%; Qwen3.5 397B scored 87.9%. Kimi K2.6 wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | Kimi K2.6 | Qwen3.5 397B | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | Kimi K2.6$0.95 input / $4 output | Qwen3.5 397B$0.6 input / $3.6 output | Qwen3.5 397B has the lower combined listed price. |
| Generation speedtokens per second | Kimi K2.6Not available | Qwen3.5 397B96 tok/s | A complete speed comparison is not available. |
| First-answer latencyseconds to first token | Kimi K2.6Not available | Qwen3.5 397B2.44 s | A complete latency comparison is not available. |
| Context windowmaximum listed tokens | Kimi K2.6256K | Qwen3.5 397B128K | Kimi K2.6 lists the larger context window. |
Benchmark Deep Dive
AgenticKimi K2.6 wins27 benchmarks
| Benchmark | Kimi K2.6 | Qwen3.5 397B | Result |
|---|---|---|---|
| Terminal-Bench 2.0Source | 66.7% | 52.5% | Kimi K2.6 leads |
| BrowseCompSource | 83.2% | 62% | Kimi K2.6 leads |
| OSWorld-VerifiedSource | 73.1% | — | Not comparable |
| ToolathlonSource | 50% | 36.3% | Kimi K2.6 leads |
| MCP AtlasSource | 55.9% | 46.1% | Kimi K2.6 leads |
| Claw-EvalSource | 62.3% | 56.8% | Kimi K2.6 leads |
| DeepSearchQASource | 92.5% | — | Not comparable |
| WideResearchSource | 80.8% | 74.0% | Kimi K2.6 leads |
| AA Agentic IndexSource | 30.3% | 19.9% | Kimi K2.6 leads |
| Tau2-TelecomSource | 95.9% | 95.6% | Kimi K2.6 leads |
| GDPval-AASource | 34.5% | 23.0% | Kimi K2.6 leads |
| GDPval-AASource | 1190 | 960 | Kimi K2.6 leads |
| APEX-Agents-AASource | 28.5% | 15.3% | Kimi K2.6 leads |
| Gert LabsSource | 56.82% | 46.76% | Kimi K2.6 leads |
| ResearchClawBenchSource | 18.0% | 14.2% | Kimi K2.6 leads |
| OSWorld 2.0Source | 4.6% | — | Not comparable |
| AA BriefcaseSource | 809 | — | Not comparable |
| AA AutomationBenchSource | 19.6% | — | Not comparable |
| AA EnterpriseOps-GymSource | 38.5% | — | Not comparable |
| AA Harvey LABSource | 0.0% | — | Not comparable |
| AA ITBenchSource | 31.2% | — | Not comparable |
| AA Tau3 BankingSource | 20.6% | — | Not comparable |
| QwenClawBenchSource | — | 51.8% | Not comparable |
| TAU3-BenchSource | — | 68.4% | Not comparable |
| VITA-BenchSource | — | 43.7% | Not comparable |
| DeepPlanningSource | — | 37.6% | Not comparable |
| MCP-TasksSource | — | 74.2% | Not comparable |
CodingKimi K2.6 wins13 benchmarks
| Benchmark | Kimi K2.6 | Qwen3.5 397B | Result |
|---|---|---|---|
| SWE-bench VerifiedSource | 80.2% | 76.2% | Kimi K2.6 leads |
| LiveCodeBenchSource | 89.6% | — | Not comparable |
| LiveCodeBench v6Source | 89.6% | 83.6% | Kimi K2.6 leads |
| SWE-bench ProSource | 58.6% | 50.9% | Kimi K2.6 leads |
| SWE MultilingualSource | 76.7% | — | Not comparable |
| SciCodeSource | 52.2% | — | Not comparable |
| Terminal-Bench 2.0Source | 66.7% | — | Not comparable |
| Vibe Code BenchSource | 37.89% | — | Not comparable |
| cursorBench31Source | 47.6% | — | Not comparable |
| AA Coding IndexSource | 61.8% | 48.2% | Kimi K2.6 leads |
| Terminal-Bench HardSource | 43.9% | 40.9% | Kimi K2.6 leads |
| AA-SciCodeSource | 53.5% | 42.0% | Kimi K2.6 leads |
| AA Terminal-Bench 2.1Source | 65.9% | — | Not comparable |
Reasoning4 benchmarks
KnowledgeQwen3.5 397B wins14 benchmarks
| Benchmark | Kimi K2.6 | Qwen3.5 397B | Result |
|---|---|---|---|
| GPQASource | 90.5% | 88.4% | Kimi K2.6 leads |
| GPQA-DSource | 90.5% | — | Not comparable |
| HLESource | 34.7% | 28.7% | Kimi K2.6 leads |
| Artificial Analysis Intelligence IndexSource | 44.2% | 33.7% | Kimi K2.6 leads |
| AA-GPQA DiamondSource | 91.1% | 89.3% | Kimi K2.6 leads |
| AA-HLESource | 35.9% | 27.3% | Kimi K2.6 leads |
| AA-Omniscience IndexSource | 6.4% | -29.8% | Kimi K2.6 leads |
| AA-Omniscience AccuracySource | 32.8% | 31.4% | Kimi K2.6 leads |
| AA-Omniscience Hallucination RateSource | 39.3% | 89.1% | Kimi K2.6 leads |
| AA Openness IndexSource | 33.3% | — | Not comparable |
| SuperGPQASource | — | 70.4% | Not comparable |
| MMLU-ProSource | — | 87.8% | Not comparable |
| MMLU-ReduxSource | — | 94.9% | Not comparable |
| C-EvalSource | — | 93% | Not comparable |
MathQwen3.5 397B wins7 benchmarks
| Benchmark | Kimi K2.6 | Qwen3.5 397B | Result |
|---|---|---|---|
| AIME26Source | 96.4% | 93.3% | Kimi K2.6 leads |
| HMMT Feb 2026Source | 92.7% | 87.9% | Kimi K2.6 leads |
| MMAnswerBenchSource | 86.0% | 80.9% | Kimi K2.6 leads |
| FrontierMath v2 (Tiers 1-3)Source | 38.966% | — | Not comparable |
| FrontierMath v2 (Tier 4)Source | 14.580% | — | Not comparable |
| HMMT Feb 2025Source | — | 94.8% | Not comparable |
| HMMT Nov 2025Source | — | 92.7% | Not comparable |
Multilingual2 benchmarks
MultimodalKimi K2.6 wins9 benchmarks
| Benchmark | Kimi K2.6 | Qwen3.5 397B | Result |
|---|---|---|---|
| MMMU-ProSource | 79.4% | 79% | Kimi K2.6 leads |
| MMMU-Pro w/ PythonSource | 80.1% | — | Not comparable |
| CharXivSource | 80.4% | 80.8% | Qwen3.5 397B leads |
| MathVisionSource | 87.4% | 88.6% | Qwen3.5 397B leads |
| V*Source | 96.9% | 95.8% | Kimi K2.6 leads |
| AA-MMMU-ProSource | 79.4% | 77.3% | Kimi K2.6 leads |
| Design Arena WebsiteSource | 1318 | — | Not comparable |
| VideoMMMUSource | — | 84.7% | Not comparable |
| ScreenSpot ProSource | — | 65.6% | Not comparable |
Frequently Asked Questions (6)
Which is better, Kimi K2.6 or Qwen3.5 397B?
Kimi K2.6 is ahead on BenchLM's provisional leaderboard, 74 to 60. The biggest single separator in this matchup is BrowseComp, where the scores are 83.2% and 62%.
Which is better for knowledge tasks, Kimi K2.6 or Qwen3.5 397B?
Qwen3.5 397B has the edge for knowledge tasks in this comparison, averaging 56.9 versus 42.2. Inside this category, AA-Omniscience Hallucination Rate is the benchmark that creates the most daylight between them.
Which is better for coding, Kimi K2.6 or Qwen3.5 397B?
Kimi K2.6 has the edge for coding in this comparison, averaging 72.6 versus 60.3. Inside this category, AA Coding Index is the benchmark that creates the most daylight between them.
Which is better for math, Kimi K2.6 or Qwen3.5 397B?
Qwen3.5 397B has the edge for math in this comparison, averaging 90.6 versus 67.1. Inside this category, MMAnswerBench is the benchmark that creates the most daylight between them.
Which is better for agentic tasks, Kimi K2.6 or Qwen3.5 397B?
Kimi K2.6 has the edge for agentic tasks in this comparison, averaging 73.5 versus 56.5. Inside this category, GDPval-AA is the benchmark that creates the most daylight between them.
Which is better for multimodal and grounded tasks, Kimi K2.6 or Qwen3.5 397B?
Kimi K2.6 has the edge for multimodal and grounded tasks in this comparison, averaging 79.8 versus 79.6. Inside this category, AA-MMMU-Pro is the benchmark that creates the most daylight between them.
Self-host vs API cost
Estimates at 50,000 req/day · 1000 tokens/req average.
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.