DeepSeek OCR 2 vs Gemma 4 31B IT vs Qwen3.5 Flash
Qwen3.5 Flash comes out ahead, 79 to 70 and 32 on our weighted score, and it is the cheaper option too.
DeepSeekDeepSeek OCR 2
32/100- ECI—
- Price$0.89 / $1.47
- Context8K
GoogleGemma 4 31B IT
70/100- ECI142.8
- Price$0.14 / $0.40
- Context262K
- Our pick
Alibaba (Qwen)Qwen3.5 Flash
79/100- ECI144.0
- Price$0.10 / $0.40
- Context1M
Qwen3.5 Flash is our pick
Qwen3.5 Flash is the better all-round choice, scoring 79/100 against Gemma 4 31B IT (70) and DeepSeek OCR 2 (32). It leads on price, 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 priceQwen3.5 FlashQwen3.5 Flash $0.175 · Gemma 4 31B IT $0.205 · DeepSeek OCR 2 $1.03 per 1M tokens (3:1 blend)
- Longest contextQwen3.5 FlashQwen3.5 Flash 1,000,000 · Gemma 4 31B IT 262,144 · DeepSeek OCR 2 8,192 tokens
- Widest inputsQwen3.5 FlashDeepSeek OCR 2: Text, Images · Gemma 4 31B IT: Text, Images · Qwen3.5 Flash: Text, Images, Video
- Self-hostingDeepSeek OCR 2 and Gemma 4 31B ITPublishes downloadable weights
| Measure | Weight | DeepSeek OCR 2 | Gemma 4 31B IT | Qwen3.5 Flash |
|---|---|---|---|---|
| Price | 50% | 49 | 83 | 86 |
| Inputs & features | 30% | 25 | 70 | 80 |
| Context window | 20% | 0 | 37 | 60 |
| Overall | 100% | 32/100 | 70/100 | 79/100 |
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 | DeepSeek OCR 2DeepSeek | Gemma 4 31B ITGoogle | Qwen3.5 FlashAlibaba (Qwen) |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | 142.8 | 144.0 (best) |
| ECI rank | — | #86 of 148 | #82 of 148 (best) |
| GPQA DiamondGraduate-level science questions | — | 75.8% | 82.3% (best) |
| FrontierMath Tiers 1–3Research-level mathematics | — | — | 18.3% |
| OTIS Mock AIME 2024–2025Competition mathematics | — | 73.3% | 84.4% (best) |
| SimpleQA VerifiedShort factual questions | — | 10.4% | 20.3% (best) |
| Price per million tokens | |||
| Input | $0.89 | $0.14 | $0.10 (best) |
| Output | $1.47 | $0.40 (best) | $0.40 (best) |
| Cached input | — | — | $0.01 |
| Blended (3:1) | $1.03 | $0.205 | $0.175 (best) |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 1 providers | Median of 30 providers | Official Alibaba API |
| Limits | |||
| Context window | 8,192 tokens | 262,144 tokens | 1,000,000 tokens (best) |
| Max output | 8,192 tokens | 32,768 tokens | 65,536 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | Yes | Yes |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | Yes |
| Reasoning | No | Yes | Yes |
| Tool calling | No | Yes | Yes |
| Structured output | No | Yes | Yes |
| Availability | |||
| Weights | Open | Open | Proprietary |
| API model ID | — | gemma-4-31b-it | qwen3.5-flash |
| API providers | 2 | 38 (best) | 8 |
| Released | Jan 27, 2026 | Apr 2, 2026 | Feb 23, 2026 |
| Knowledge cutoff | — | — | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- DeepSeek OCR 2$11.83
- Gemma 4 31B IT$2.20
- Qwen3.5 Flash$1.80
Which should you choose?
Which is better: DeepSeek OCR 2, Gemma 4 31B IT or Qwen3.5 Flash?
Qwen3.5 Flash is the better all-round choice, scoring 79/100 against Gemma 4 31B IT (70) and DeepSeek OCR 2 (32). It leads on price, 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, DeepSeek OCR 2, Gemma 4 31B IT or Qwen3.5 Flash?
Qwen3.5 Flash is cheaper at $0.10 input / $0.40 output per million tokens (official Alibaba API price). Gemma 4 31B IT costs $0.14 input / $0.40 output per million tokens (median across 30 API providers); DeepSeek OCR 2 costs $0.89 input / $1.47 output per million tokens (median across 1 API provider). At a typical mix of three input tokens to one output token, that is $0.175 per million tokens for Qwen3.5 Flash versus $0.205 for Gemma 4 31B IT (1.2× as much) and $1.03 for DeepSeek OCR 2 (5.9× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. DeepSeek OCR 2 has not been scored yet, Gemma 4 31B IT has an ECI of 142.8 and Qwen3.5 Flash has an ECI of 144.0.
Which is better for coding?
There are no published SWE-bench Verified results for DeepSeek OCR 2, Gemma 4 31B IT and Qwen3.5 Flash yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. Note that DeepSeek OCR 2 does not support tool calling, which most coding agents need.
Which has the bigger context window?
Qwen3.5 Flash has the largest context window at 1,000,000 tokens, against 262,144 for Gemma 4 31B IT and 8,192 for DeepSeek OCR 2. Maximum output per response: DeepSeek OCR 2 up to 8,192, Gemma 4 31B IT up to 32,768, Qwen3.5 Flash up to 65,536 tokens.
Which can read images, PDFs, audio or video?
DeepSeek OCR 2 accepts text and images; Gemma 4 31B IT accepts text and images; Qwen3.5 Flash accepts text, images and video. Qwen3.5 Flash handles the widest range of inputs.
Are any of these open source?
DeepSeek OCR 2 and Gemma 4 31B IT publishes its weights and can be self-hosted; Qwen3.5 Flash is proprietary.
Which is newer?
Gemma 4 31B IT is the newest, released Apr 2, 2026. Qwen3.5 Flash came out Feb 23, 2026; DeepSeek OCR 2 came out Jan 27, 2026.
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.
