Qwen-MT Plus vs Qwen2.5-VL 72B Instruct vs Sonar Deep Research
Qwen2.5-VL 72B Instruct comes out ahead, 30 to 20 and 12 on our weighted score, though Sonar Deep Research is 17% cheaper per token.
Alibaba (Qwen)Qwen-MT Plus
12/100- ECI—
- Price$2.46 / $7.37
- Context16K
- Our pickAlibaba (Qwen)
Qwen2.5-VL 72B Instruct
30/100- ECI—
- Price$2.80 / $8.40
- Context131K
PerplexitySonar Deep Research
20/100- ECI—
- Price$2.00 / $8.00
- Context128K
Qwen2.5-VL 72B Instruct is our pick
Qwen2.5-VL 72B Instruct is the better all-round choice, scoring 30/100 against Sonar Deep Research (20) and Qwen-MT Plus (12). It leads on inputs & features. 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 priceSonar Deep ResearchSonar Deep Research $3.50 · Qwen-MT Plus $3.69 · Qwen2.5-VL 72B Instruct $4.20 per 1M tokens (3:1 blend)
- Longest contextQwen2.5-VL 72B InstructQwen2.5-VL 72B Instruct 131,072 · Sonar Deep Research 128,000 · Qwen-MT Plus 16,384 tokens
- Widest inputsQwen2.5-VL 72B InstructQwen-MT Plus: Text · Qwen2.5-VL 72B Instruct: Text, Images · Sonar Deep Research: Text
- Self-hostingQwen2.5-VL 72B InstructPublishes downloadable weights
| Measure | Weight | Qwen-MT Plus | Qwen2.5-VL 72B Instruct | Sonar Deep Research |
|---|---|---|---|---|
| Price | 50% | 23 | 20 | 24 |
| Inputs & features | 30% | 0 | 50 | 10 |
| Context window | 20% | 0 | 24 | 24 |
| Overall | 100% | 12/100 | 30/100 | 20/100 |
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 | Qwen-MT PlusAlibaba (Qwen) | Qwen2.5-VL 72B InstructAlibaba (Qwen) | Sonar Deep ResearchPerplexity |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $2.46 | $2.80 | $2.00 (best) |
| Output | $7.37 (best) | $8.40 | $8.00 |
| Cached input | — | — | — |
| Blended (3:1) | $3.69 | $4.20 | $3.50 (best) |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Alibaba API | Official Alibaba API | Official Perplexity API |
| Limits | |||
| Context window | 16,384 tokens | 131,072 tokens (best) | 128,000 tokens |
| Max output | 8,192 tokens | 8,192 tokens | 32,768 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | Yes | No |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | No | No | Yesminimal · low · medium · high |
| Tool calling | No | Yes | No |
| Structured output | No | No | No |
| Availability | |||
| Weights | Proprietary | Open | Proprietary |
| API model ID | qwen-mt-plus | qwen2-5-vl-72b-instruct | sonar-deep-research |
| API providers | 1 | 1 | 3 (best) |
| Released | Jan 2025 | Sep 2024 | Feb 1, 2025 |
| Knowledge cutoff | Apr 2024 | Apr 2024 | Jan 2025 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Qwen-MT Plus$39.34
- Qwen2.5-VL 72B Instruct$44.80
- Sonar Deep Research$36.00
Which should you choose?
Which is better: Qwen-MT Plus, Qwen2.5-VL 72B Instruct or Sonar Deep Research?
Qwen2.5-VL 72B Instruct is the better all-round choice, scoring 30/100 against Sonar Deep Research (20) and Qwen-MT Plus (12). It leads on inputs & features. 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, Qwen-MT Plus, Qwen2.5-VL 72B Instruct or Sonar Deep Research?
Sonar Deep Research is cheaper at $2.00 input / $8.00 output per million tokens (official Perplexity API price). Qwen-MT Plus costs $2.46 input / $7.37 output per million tokens (official Alibaba API price); Qwen2.5-VL 72B Instruct costs $2.80 input / $8.40 output per million tokens (official Alibaba API price). At a typical mix of three input tokens to one output token, that is $3.50 per million tokens for Sonar Deep Research versus $3.69 for Qwen-MT Plus (1.1× as much) and $4.20 for Qwen2.5-VL 72B Instruct (1.2× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Qwen-MT Plus has not been scored yet, Qwen2.5-VL 72B Instruct has not been scored yet and Sonar Deep Research has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Qwen-MT Plus, Qwen2.5-VL 72B Instruct and Sonar Deep Research yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. Note that Qwen-MT Plus and Sonar Deep Research does not support tool calling, which most coding agents need.
Which has the bigger context window?
Qwen2.5-VL 72B Instruct has the largest context window at 131,072 tokens, against 128,000 for Sonar Deep Research and 16,384 for Qwen-MT Plus. Maximum output per response: Qwen-MT Plus up to 8,192, Qwen2.5-VL 72B Instruct up to 8,192, Sonar Deep Research up to 32,768 tokens.
Which can read images, PDFs, audio or video?
Qwen-MT Plus accepts text; Qwen2.5-VL 72B Instruct accepts text and images; Sonar Deep Research accepts text. Qwen2.5-VL 72B Instruct handles the widest range of inputs.
Are any of these open source?
Qwen2.5-VL 72B Instruct publishes its weights and can be self-hosted; Qwen-MT Plus and Sonar Deep Research is proprietary.
Which is newer?
Sonar Deep Research is the newest, released Feb 1, 2025. Qwen-MT Plus came out Jan 2025; Qwen2.5-VL 72B Instruct came out Sep 2024. Knowledge cutoff: Qwen-MT Plus Apr 2024, Qwen2.5-VL 72B Instruct Apr 2024, Sonar Deep Research Jan 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.
