Mistral Large 3 vs Nova 2 Lite vs Qwen3 VL 235B A22B Thinking
Nova 2 Lite comes out ahead, 62 to 50 and 48 on our weighted score, though Mistral Large 3 is 20% cheaper per token.
Mistral AIMistral Large 3
50/100- ECI—
- Price$0.50 / $1.50
- Context262K
- Our pick
AmazonNova 2 Lite
62/100- ECI—
- Price$0.33 / $2.75
- Context1M
Alibaba (Qwen)Qwen3 VL 235B A22B Thinking
48/100- ECI—
- Price$0.40 / $4.00
- Context131K
Nova 2 Lite is our pick
Nova 2 Lite is the better all-round choice, scoring 62/100 against Mistral Large 3 (50) and Qwen3 VL 235B A22B Thinking (48). It leads on inputs & features and context window. Mistral Large 3 wins on price. 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 priceMistral Large 3Mistral Large 3 $0.75 · Nova 2 Lite $0.935 · Qwen3 VL 235B A22B Thinking $1.30 per 1M tokens (3:1 blend)
- Longest contextNova 2 LiteNova 2 Lite 1,000,000 · Mistral Large 3 262,144 · Qwen3 VL 235B A22B Thinking 131,072 tokens
- Widest inputsNova 2 LiteMistral Large 3: Text, Images · Nova 2 Lite: Text, Images, PDFs, Video · Qwen3 VL 235B A22B Thinking: Text, Images
- Self-hostingMistral Large 3 and Qwen3 VL 235B A22B ThinkingPublishes downloadable weights
| Measure | Weight | Mistral Large 3 | Nova 2 Lite | Qwen3 VL 235B A22B Thinking |
|---|---|---|---|---|
| Price | 50% | 56 | 51 | 44 |
| Inputs & features | 30% | 50 | 80 | 70 |
| Context window | 20% | 37 | 60 | 24 |
| Overall | 100% | 50/100 | 62/100 | 48/100 |
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 | Mistral Large 3Mistral AI | Nova 2 LiteAmazon | Qwen3 VL 235B A22B ThinkingAlibaba (Qwen) |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $0.50 | $0.33 (best) | $0.40 |
| Output | $1.50 (best) | $2.75 | $4.00 |
| Cached input | $0.05 (best) | $0.083 | — |
| Blended (3:1) | $0.75 (best) | $0.935 | $1.30 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Mistral API | Official Amazon Bedrock API | Median of 9 providers |
| Limits | |||
| Context window | 262,144 tokens | 1,000,000 tokens (best) | 131,072 tokens |
| Max output | 262,144 tokens (best) | 65,536 tokens | 32,768 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | Yes | Yes |
| PDFs | No | Yes | No |
| Audio | No | No | No |
| Video | No | Yes | No |
| Reasoning | No | Yeslow · medium · high | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | Yes |
| Availability | |||
| Weights | Open | Proprietary | Open |
| API model ID | mistral-large-2512 | amazon.nova-2-lite-v1:0 | — |
| API providers | 13 (best) | 2 | 9 |
| Released | Dec 2, 2025 | Dec 2, 2025 | Sep 23, 2025 |
| Knowledge cutoff | Nov 2024 | Oct 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.
- Mistral Large 3$8.00
- Nova 2 Lite$8.80
- Qwen3 VL 235B A22B Thinking$12.00
Which should you choose?
Which is better: Mistral Large 3, Nova 2 Lite or Qwen3 VL 235B A22B Thinking?
Nova 2 Lite is the better all-round choice, scoring 62/100 against Mistral Large 3 (50) and Qwen3 VL 235B A22B Thinking (48). It leads on inputs & features and context window. Mistral Large 3 wins on price. 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, Mistral Large 3, Nova 2 Lite or Qwen3 VL 235B A22B Thinking?
Mistral Large 3 is cheaper at $0.50 input / $1.50 output per million tokens (official Mistral API price). Nova 2 Lite costs $0.33 input / $2.75 output per million tokens (official Amazon Bedrock API price); Qwen3 VL 235B A22B Thinking costs $0.40 input / $4.00 output per million tokens (median across 9 API providers). At a typical mix of three input tokens to one output token, that is $0.75 per million tokens for Mistral Large 3 versus $0.935 for Nova 2 Lite (1.2× as much) and $1.30 for Qwen3 VL 235B A22B Thinking (1.7× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Mistral Large 3 has not been scored yet, Nova 2 Lite has not been scored yet and Qwen3 VL 235B A22B Thinking has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Mistral Large 3, Nova 2 Lite and Qwen3 VL 235B A22B Thinking 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?
Nova 2 Lite has the largest context window at 1,000,000 tokens, against 262,144 for Mistral Large 3 and 131,072 for Qwen3 VL 235B A22B Thinking. Maximum output per response: Mistral Large 3 up to 262,144, Nova 2 Lite up to 65,536, Qwen3 VL 235B A22B Thinking up to 32,768 tokens.
Which can read images, PDFs, audio or video?
Mistral Large 3 accepts text and images; Nova 2 Lite accepts text, images, PDFs and video; Qwen3 VL 235B A22B Thinking accepts text and images. Nova 2 Lite handles the widest range of inputs.
Are any of these open source?
Mistral Large 3 and Qwen3 VL 235B A22B Thinking publishes its weights and can be self-hosted; Nova 2 Lite is proprietary.
Which is newer?
Mistral Large 3 is the newest, released Dec 2, 2025. Nova 2 Lite came out Dec 2, 2025; Qwen3 VL 235B A22B Thinking came out Sep 23, 2025. Knowledge cutoff: Mistral Large 3 Nov 2024, Nova 2 Lite Oct 2025, Qwen3 VL 235B A22B Thinking 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.
