Apertus 70B vs Kimi K2.5 vs Qwen3.6 Plus
Qwen3.6 Plus comes out ahead, 59 to 50 and 32 on our weighted score, and it is the cheaper option too.
Swiss AI
Apertus 70B
32/100- ECI—
- Price$0.82 / $2.42
- Context66K
Moonshot AIKimi K2.5
50/100- ECI148.1
- Price$0.60 / $3.00
- Context262K
- Our pick
Alibaba (Qwen)Qwen3.6 Plus
59/100- ECI147.8
- Price$0.50 / $3.00
- Context1M
Qwen3.6 Plus is our pick
Qwen3.6 Plus is the better all-round choice, scoring 59/100 against Kimi K2.5 (50) and Apertus 70B (32). It leads on inputs & features and context window. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
- CapabilityNot enough dataNo independent benchmark covers every model here yet
- Lowest priceQwen3.6 PlusQwen3.6 Plus $1.13 · Kimi K2.5 $1.20 · Apertus 70B $1.22 per 1M tokens (3:1 blend)
- Longest contextQwen3.6 PlusQwen3.6 Plus 1,000,000 · Kimi K2.5 262,144 · Apertus 70B 65,536 tokens
- Widest inputsKimi K2.5 and Qwen3.6 PlusApertus 70B: Text · Kimi K2.5: Text, Images, Video · Qwen3.6 Plus: Text, Images, Video
- Self-hostingApertus 70B and Kimi K2.5Publishes downloadable weights (Apache-2.0)
| Measure | Weight | Apertus 70B | Kimi K2.5 | Qwen3.6 Plus |
|---|---|---|---|---|
| Price | 50% | 44 | 44 | 45 |
| Inputs & features | 30% | 25 | 70 | 80 |
| Context window | 20% | 12 | 37 | 60 |
| Overall | 100% | 32/100 | 50/100 | 59/100 |
Left out because at least one model lacks the data: capability. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Apertus 70BSwiss AI | Kimi K2.5Moonshot AI | Qwen3.6 PlusAlibaba (Qwen) |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | 148.1 (best) | 147.8 |
| ECI rank | — | #68 of 165 (best) | #69 of 165 |
| GPQA DiamondGraduate-level science questions | — | 87.6% | 88.4% (best) |
| FrontierMath Tiers 1–3Research-level mathematics | — | — | 38.3% |
| OTIS Mock AIME 2024–2025Competition mathematics | — | 92.2% | 93.3% (best) |
| SWE-bench VerifiedFixing real GitHub issues | — | 73.8% (best) | 57.9% |
| SimpleQA VerifiedShort factual questions | — | 34.3% | 44.1% (best) |
| Price per million tokens | |||
| Input | $0.82 | $0.60 | $0.50 (best) |
| Output | $2.42 (best) | $3.00 | $3.00 |
| Cached input | — | — | $0.05 |
| Blended (3:1) | $1.22 | $1.20 | $1.13 (best) |
| Long-context rate | Same rate | Same rate | Over 256K: $2.00 / $6.00 |
| Price source | Median of 3 providers | Median of 33 providers | Official Alibaba API |
| Limits | |||
| Context window | 65,536 tokens | 262,144 tokens | 1,000,000 tokens (best) |
| Max output | 8,192 tokens | 262,144 tokens (best) | 65,536 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | Yes | Yes |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | Yes | Yes |
| Reasoning | No | Yes | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | Yes |
| Availability | |||
| Weights | OpenApache-2.0 | Open | Proprietary |
| API model ID | — | — | qwen3.6-plus |
| API providers | 3 | 33 (best) | 22 |
| Released | Sep 2, 2025 | Jan 27, 2026 | Apr 2, 2026 |
| Knowledge cutoff | Sep 2025 | Jan 2025 | Apr 2025 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Apertus 70B$13.04
- Kimi K2.5$12.00
- Qwen3.6 Plus$11.00
Which should you choose?
Which is better: Apertus 70B, Kimi K2.5 or Qwen3.6 Plus?
Qwen3.6 Plus is the better all-round choice, scoring 59/100 against Kimi K2.5 (50) and Apertus 70B (32). It leads on inputs & features and context window. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, Apertus 70B, Kimi K2.5 or Qwen3.6 Plus?
Qwen3.6 Plus is cheaper at $0.50 input / $3.00 output per million tokens (official Alibaba API price). Kimi K2.5 costs $0.60 input / $3.00 output per million tokens (median across 33 API providers); Apertus 70B costs $0.82 input / $2.42 output per million tokens (median across 3 API providers). At a typical mix of three input tokens to one output token, that is $1.13 per million tokens for Qwen3.6 Plus versus $1.20 for Kimi K2.5 (1.1× as much) and $1.22 for Apertus 70B (1.1× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Apertus 70B has not been scored yet, Kimi K2.5 has an ECI of 148.1 and Qwen3.6 Plus has an ECI of 147.8.
Which is better for coding?
There are no published SWE-bench Verified results for Apertus 70B yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. All three support tool calling for agent workflows.
Which has the bigger context window?
Qwen3.6 Plus has the largest context window at 1,000,000 tokens, against 262,144 for Kimi K2.5 and 65,536 for Apertus 70B. Maximum output per response: Apertus 70B up to 8,192, Kimi K2.5 up to 262,144, Qwen3.6 Plus up to 65,536 tokens.
Which can read images, PDFs, audio or video?
Apertus 70B accepts text; Kimi K2.5 accepts text, images and video; Qwen3.6 Plus accepts text, images and video. Kimi K2.5 handles the widest range of inputs.
Are any of these open source?
Apertus 70B and Kimi K2.5 publishes its weights (Apache-2.0) and can be self-hosted; Qwen3.6 Plus is proprietary.
Which is newer?
Qwen3.6 Plus is the newest, released Apr 2, 2026. Kimi K2.5 came out Jan 27, 2026; Apertus 70B came out Sep 2, 2025. Knowledge cutoff: Apertus 70B Sep 2025, Kimi K2.5 Jan 2025, Qwen3.6 Plus Apr 2025.
How do you decide the winner?
Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Nov 2, 2026.
