Granite-4.0-H-Micro vs PaddleOCR VL vs Voxtral Mini 3B 2507
Too close to call on our weighted score (Granite-4.0-H-Micro 62, PaddleOCR VL 62, Voxtral Mini 3B 2507 61). The right pick depends on what you value most.
IBMGranite-4.0-H-Micro
62/100- ECI—
- Price$0.018 / $0.115
- Context131K
PaddlePaddle
PaddleOCR VL
62/100- ECI—
- Price$0.02 / $0.02
- Context131K
Mistral AIVoxtral Mini 3B 2507
61/100- ECI—
- Price$0.04 / $0.04
- Context33K
Too close to call
It is close. Our weighted score puts them within a point (Granite-4.0-H-Micro 62/100, PaddleOCR VL 62/100, Voxtral Mini 3B 2507 61/100), so choose by what matters most for your work: PaddleOCR VL 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 pricePaddleOCR VLPaddleOCR VL $0.02 · Voxtral Mini 3B 2507 $0.04 · Granite-4.0-H-Micro $0.042 per 1M tokens (3:1 blend)
- Longest contextGranite-4.0-H-Micro and PaddleOCR VLGranite-4.0-H-Micro 131,072 · PaddleOCR VL 131,072 · Voxtral Mini 3B 2507 32,768 tokens
- Widest inputsPaddleOCR VL and Voxtral Mini 3B 2507Granite-4.0-H-Micro: Text · PaddleOCR VL: Text, Images · Voxtral Mini 3B 2507: Text, Audio
- Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
| Measure | Weight | Granite-4.0-H-Micro | PaddleOCR VL | Voxtral Mini 3B 2507 |
|---|---|---|---|---|
| Price | 50% | 100 | 100 | 100 |
| Inputs & features | 30% | 25 | 25 | 35 |
| Context window | 20% | 25 | 25 | 0 |
| Overall | 100% | 62/100 | 62/100 | 61/100 |
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 | Granite-4.0-H-MicroIBM | PaddleOCR VLPaddlePaddle | Voxtral Mini 3B 2507Mistral AI |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $0.018 (best) | $0.02 | $0.04 |
| Output | $0.115 | $0.02 (best) | $0.04 |
| Cached input | — | — | — |
| Blended (3:1) | $0.042 | $0.02 (best) | $0.04 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 2 providers | Median of 1 providers | Median of 2 providers |
| Limits | |||
| Context window | 131,072 tokens (best) | 131,072 tokens (best) | 32,768 tokens |
| Max output | 131,072 tokens (best) | 16,384 tokens | 32,768 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | Yes | No |
| PDFs | No | No | No |
| Audio | No | No | Yes |
| Video | No | No | No |
| Reasoning | No | No | No |
| Tool calling | Yes | No | Yes |
| Structured output | No | No | No |
| Availability | |||
| Weights | Open | OpenApache-2.0 | OpenApache 2.0 |
| API model ID | — | — | — |
| API providers | 2 | 1 | 3 (best) |
| Released | Oct 2, 2025 | Oct 16, 2025 | Jul 15, 2025 |
| 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.
- Granite-4.0-H-Micro$0.405
- PaddleOCR VL$0.24
- Voxtral Mini 3B 2507$0.48
Which should you choose?
Which is better: Granite-4.0-H-Micro, PaddleOCR VL or Voxtral Mini 3B 2507?
It is close. Our weighted score puts them within a point (Granite-4.0-H-Micro 62/100, PaddleOCR VL 62/100, Voxtral Mini 3B 2507 61/100), so choose by what matters most for your work: PaddleOCR VL 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, Granite-4.0-H-Micro, PaddleOCR VL or Voxtral Mini 3B 2507?
PaddleOCR VL is cheaper at $0.02 input / $0.02 output per million tokens (median across 1 API provider). Voxtral Mini 3B 2507 costs $0.04 input / $0.04 output per million tokens (median across 2 API providers); Granite-4.0-H-Micro costs $0.018 input / $0.115 output per million tokens (median across 2 API providers). At a typical mix of three input tokens to one output token, that is $0.02 per million tokens for PaddleOCR VL versus $0.04 for Voxtral Mini 3B 2507 (2× as much) and $0.042 for Granite-4.0-H-Micro (2.1× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Granite-4.0-H-Micro has not been scored yet, PaddleOCR VL has not been scored yet and Voxtral Mini 3B 2507 has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Granite-4.0-H-Micro, PaddleOCR VL and Voxtral Mini 3B 2507 yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. Note that PaddleOCR VL does not support tool calling, which most coding agents need.
Which has the bigger context window?
Granite-4.0-H-Micro and PaddleOCR VL have the largest context windows (131,072 and 131,072 tokens), against 32,768 for Voxtral Mini 3B 2507. Maximum output per response: Granite-4.0-H-Micro up to 131,072, PaddleOCR VL up to 16,384, Voxtral Mini 3B 2507 up to 32,768 tokens.
Which can read images, PDFs, audio or video?
Granite-4.0-H-Micro accepts text; PaddleOCR VL accepts text and images; Voxtral Mini 3B 2507 accepts text and audio. PaddleOCR VL handles the widest range of inputs.
Are any of these open source?
Yes, all three publish their weights (Apache-2.0 and Apache 2.0), so you can self-host them.
Which is newer?
PaddleOCR VL is the newest, released Oct 16, 2025. Granite-4.0-H-Micro came out Oct 2, 2025; Voxtral Mini 3B 2507 came out Jul 15, 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 Nov 2, 2026.
