Devstral 2 vs Qwen3.5 27B vs Nova 2 Lite
Too close to call on our weighted score (Nova 2 Lite 62, Qwen3.5 27B 61, Devstral 2 42). The right pick depends on what you value most.
Mistral AIDevstral 2
42/100- ECI—
- Price$0.40 / $2.00
- Context262K
Alibaba (Qwen)Qwen3.5 27B
61/100- ECI—
- Price$0.30 / $2.40
- Context262K
AmazonNova 2 Lite
62/100- ECI—
- Price$0.33 / $2.75
- Context1M
Too close to call
It is close. Our weighted score puts them within a point (Nova 2 Lite 62/100, Qwen3.5 27B 61/100, Devstral 2 42/100), so choose by what matters most for your work: Devstral 2 on price and Nova 2 Lite for long inputs. 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 priceDevstral 2Devstral 2 $0.80 · Qwen3.5 27B $0.825 · Nova 2 Lite $0.935 per 1M tokens (3:1 blend)
- Longest contextNova 2 LiteNova 2 Lite 1,000,000 · Devstral 2 262,144 · Qwen3.5 27B 262,144 tokens
- Widest inputsQwen3.5 27B and Nova 2 LiteDevstral 2: Text · Qwen3.5 27B: Text, Images, Audio, Video · Nova 2 Lite: Text, Images, PDFs, Video
- Self-hostingDevstral 2 and Qwen3.5 27BPublishes downloadable weights
| Measure | Weight | Devstral 2 | Qwen3.5 27B | Nova 2 Lite |
|---|---|---|---|---|
| Price | 50% | 54 | 54 | 51 |
| Inputs & features | 30% | 25 | 90 | 80 |
| Context window | 20% | 37 | 37 | 60 |
| Overall | 100% | 42/100 | 61/100 | 62/100 |
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 | Devstral 2Mistral AI | Qwen3.5 27BAlibaba (Qwen) | Nova 2 LiteAmazon |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $0.40 | $0.30 (best) | $0.33 |
| Output | $2.00 (best) | $2.40 | $2.75 |
| Cached input | — | — | $0.083 |
| Blended (3:1) | $0.80 (best) | $0.825 | $0.935 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Mistral API | Official Alibaba API | Official Amazon Bedrock API |
| Limits | |||
| Context window | 262,144 tokens | 262,144 tokens | 1,000,000 tokens (best) |
| Max output | 262,144 tokens (best) | 65,536 tokens | 65,536 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | Yes | Yes |
| PDFs | No | No | Yes |
| Audio | No | Yes | No |
| Video | No | Yes | Yes |
| Reasoning | No | Yes | Yeslow · medium · high |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | Yes | No |
| Availability | |||
| Weights | Open | Open | Proprietary |
| API model ID | devstral-2512 | qwen3.5-27b | amazon.nova-2-lite-v1:0 |
| API providers | 14 | 16 (best) | 2 |
| Released | Dec 9, 2025 | Feb 23, 2026 | Dec 2, 2025 |
| Knowledge cutoff | Dec 2025 | — | Oct 2025 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Devstral 2$8.00
- Qwen3.5 27B$7.80
- Nova 2 Lite$8.80
Which should you choose?
Which is better: Devstral 2, Qwen3.5 27B or Nova 2 Lite?
It is close. Our weighted score puts them within a point (Nova 2 Lite 62/100, Qwen3.5 27B 61/100, Devstral 2 42/100), so choose by what matters most for your work: Devstral 2 on price and Nova 2 Lite for long inputs. 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, Devstral 2, Qwen3.5 27B or Nova 2 Lite?
Devstral 2 is cheaper at $0.40 input / $2.00 output per million tokens (official Mistral API price). Qwen3.5 27B costs $0.30 input / $2.40 output per million tokens (official Alibaba API price); Nova 2 Lite costs $0.33 input / $2.75 output per million tokens (official Amazon Bedrock API price). At a typical mix of three input tokens to one output token, that is $0.80 per million tokens for Devstral 2 versus $0.825 for Qwen3.5 27B (1× as much) and $0.935 for Nova 2 Lite (1.2× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Devstral 2 has not been scored yet, Qwen3.5 27B has not been scored yet and Nova 2 Lite has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Devstral 2, Qwen3.5 27B and Nova 2 Lite 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 Devstral 2 and 262,144 for Qwen3.5 27B. Maximum output per response: Devstral 2 up to 262,144, Qwen3.5 27B up to 65,536, Nova 2 Lite up to 65,536 tokens.
Which can read images, PDFs, audio or video?
Devstral 2 accepts text; Qwen3.5 27B accepts text, images, audio and video; Nova 2 Lite accepts text, images, PDFs and video. Qwen3.5 27B handles the widest range of inputs.
Are any of these open source?
Devstral 2 and Qwen3.5 27B publishes its weights and can be self-hosted; Nova 2 Lite is proprietary.
Which is newer?
Qwen3.5 27B is the newest, released Feb 23, 2026. Devstral 2 came out Dec 9, 2025; Nova 2 Lite came out Dec 2, 2025. Knowledge cutoff: Devstral 2 Dec 2025, Nova 2 Lite Oct 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.
