Apertus 70B vs MiniMax-M2 vs Qwen3 VL 235B A22B Instruct
Qwen3 VL 235B A22B Instruct comes out ahead, 53 to 48 and 33 on our weighted score, though MiniMax-M2 is 14% cheaper per token.
Swiss AI
Apertus 70B
33/100- ECI—
- Price$0.82 / $2.42
- Context66K
MiniMaxMiniMax-M2
48/100- ECI—
- Price$0.30 / $1.20
- Context205K
- Our pick
Alibaba (Qwen)Qwen3 VL 235B A22B Instruct
53/100- ECI—
- Price$0.30 / $1.55
- Context131K
Qwen3 VL 235B A22B Instruct is our pick
Qwen3 VL 235B A22B Instruct is the better all-round choice, scoring 53/100 against MiniMax-M2 (48) and Apertus 70B (33). It leads on inputs & features. MiniMax-M2 wins on price 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 priceMiniMax-M2MiniMax-M2 $0.525 · Qwen3 VL 235B A22B Instruct $0.613 · Apertus 70B $1.22 per 1M tokens (3:1 blend)
- Longest contextMiniMax-M2MiniMax-M2 204,800 · Qwen3 VL 235B A22B Instruct 131,072 · Apertus 70B 65,536 tokens
- Widest inputsQwen3 VL 235B A22B InstructApertus 70B: Text · MiniMax-M2: Text · Qwen3 VL 235B A22B Instruct: Text, Images
- Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
| Measure | Weight | Apertus 70B | MiniMax-M2 | Qwen3 VL 235B A22B Instruct |
|---|---|---|---|---|
| Price | 50% | 46 | 63 | 60 |
| Inputs & features | 30% | 25 | 35 | 60 |
| Context window | 20% | 12 | 32 | 24 |
| Overall | 100% | 33/100 | 48/100 | 53/100 |
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 | MiniMax-M2MiniMax | Qwen3 VL 235B A22B InstructAlibaba (Qwen) |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $0.82 | $0.30 (best) | $0.30 (best) |
| Output | $2.42 | $1.20 (best) | $1.55 |
| Cached input | — | — | — |
| Blended (3:1) | $1.22 | $0.525 (best) | $0.613 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 3 providers | Official MiniMax (minimax.io) API | Median of 12 providers |
| Limits | |||
| Context window | 65,536 tokens | 204,800 tokens (best) | 131,072 tokens |
| Max output | 8,192 tokens | 131,072 tokens (best) | 32,768 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | No | Yes |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | No | Yes | No |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | Yes |
| Availability | |||
| Weights | OpenApache-2.0 | Open | Open |
| API model ID | — | MiniMax-M2 | — |
| API providers | 3 | 13 (best) | 12 |
| Released | Sep 2, 2025 | Oct 27, 2025 | Sep 23, 2025 |
| Knowledge cutoff | Sep 2025 | — | Mar 31, 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
- MiniMax-M2$5.40
- Qwen3 VL 235B A22B Instruct$6.10
Which should you choose?
Which is better: Apertus 70B, MiniMax-M2 or Qwen3 VL 235B A22B Instruct?
Qwen3 VL 235B A22B Instruct is the better all-round choice, scoring 53/100 against MiniMax-M2 (48) and Apertus 70B (33). It leads on inputs & features. MiniMax-M2 wins on price 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, MiniMax-M2 or Qwen3 VL 235B A22B Instruct?
MiniMax-M2 is cheaper at $0.30 input / $1.20 output per million tokens (official MiniMax (minimax.io) API price). Qwen3 VL 235B A22B Instruct costs $0.30 input / $1.55 output per million tokens (median across 12 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 $0.525 per million tokens for MiniMax-M2 versus $0.613 for Qwen3 VL 235B A22B Instruct (1.2× as much) and $1.22 for Apertus 70B (2.3× 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, MiniMax-M2 has not been scored yet and Qwen3 VL 235B A22B Instruct has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Apertus 70B, MiniMax-M2 and Qwen3 VL 235B A22B Instruct 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?
MiniMax-M2 has the largest context window at 204,800 tokens, against 131,072 for Qwen3 VL 235B A22B Instruct and 65,536 for Apertus 70B. Maximum output per response: Apertus 70B up to 8,192, MiniMax-M2 up to 131,072, Qwen3 VL 235B A22B Instruct up to 32,768 tokens.
Which can read images, PDFs, audio or video?
Apertus 70B accepts text; MiniMax-M2 accepts text; Qwen3 VL 235B A22B Instruct accepts text and images. Qwen3 VL 235B A22B Instruct handles the widest range of inputs.
Are any of these open source?
Yes, all three publish their weights (Apache-2.0), so you can self-host them.
Which is newer?
MiniMax-M2 is the newest, released Oct 27, 2025. Qwen3 VL 235B A22B Instruct came out Sep 23, 2025; Apertus 70B came out Sep 2, 2025. Knowledge cutoff: Apertus 70B Sep 2025, Qwen3 VL 235B A22B Instruct Mar 31, 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 Oct 4, 2026.
