Apertus 8B vs Devstral Small 2 vs Granite-4.0-H-Small
Devstral Small 2 comes out ahead, 67 to 63 and 56 on our weighted score, though Granite-4.0-H-Small is 24% cheaper per token.
Swiss AI
Apertus 8B
56/100- ECI—
- Price$0.10 / $0.20
- Context66K
- Our pick
Mistral AIDevstral Small 2
67/100- ECI—
- Price$0.10 / $0.30
- Context262K
IBMGranite-4.0-H-Small
63/100- ECI—
- Price$0.064 / $0.265
- Context131K
Devstral Small 2 is our pick
Devstral Small 2 is the better all-round choice, scoring 67/100 against Granite-4.0-H-Small (63) and Apertus 8B (56). It leads on inputs & features and context window. Granite-4.0-H-Small 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 priceGranite-4.0-H-SmallGranite-4.0-H-Small $0.114 · Apertus 8B $0.125 · Devstral Small 2 $0.15 per 1M tokens (3:1 blend)
- Longest contextDevstral Small 2Devstral Small 2 262,144 · Granite-4.0-H-Small 131,072 · Apertus 8B 65,536 tokens
- Widest inputsDevstral Small 2Apertus 8B: Text · Devstral Small 2: Text, Images · Granite-4.0-H-Small: Text
- Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
| Measure | Weight | Apertus 8B | Devstral Small 2 | Granite-4.0-H-Small |
|---|---|---|---|---|
| Price | 50% | 93 | 89 | 95 |
| Inputs & features | 30% | 25 | 50 | 35 |
| Context window | 20% | 12 | 37 | 24 |
| Overall | 100% | 56/100 | 67/100 | 63/100 |
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 | Apertus 8BSwiss AI | Devstral Small 2Mistral AI | Granite-4.0-H-SmallIBM |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $0.10 | $0.10 | $0.064 (best) |
| Output | $0.20 (best) | $0.30 | $0.265 |
| Cached input | — | — | — |
| Blended (3:1) | $0.125 | $0.15 | $0.114 (best) |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 1 providers | Median of 1 providers | Official watsonx.ai API |
| Limits | |||
| Context window | 65,536 tokens | 262,144 tokens (best) | 131,072 tokens |
| Max output | 8,192 tokens | 262,144 tokens (best) | 131,072 tokens |
| 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 | No |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | Yes |
| Availability | |||
| Weights | OpenApache-2.0 | OpenApache-2.0 | Open |
| API model ID | — | — | ibm/granite-4-h-small |
| API providers | 1 | 1 | 1 |
| Released | Sep 2, 2025 | Dec 9, 2025 | Oct 2, 2025 |
| Knowledge cutoff | Sep 2025 | Dec 2025 | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Apertus 8B$1.40
- Devstral Small 2$1.60
- Granite-4.0-H-Small$1.17
Which should you choose?
Which is better: Apertus 8B, Devstral Small 2 or Granite-4.0-H-Small?
Devstral Small 2 is the better all-round choice, scoring 67/100 against Granite-4.0-H-Small (63) and Apertus 8B (56). It leads on inputs & features and context window. Granite-4.0-H-Small 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, Apertus 8B, Devstral Small 2 or Granite-4.0-H-Small?
Granite-4.0-H-Small is cheaper at $0.064 input / $0.265 output per million tokens (official watsonx.ai API price). Apertus 8B costs $0.10 input / $0.20 output per million tokens (median across 1 API provider); Devstral Small 2 costs $0.10 input / $0.30 output per million tokens (median across 1 API provider). At a typical mix of three input tokens to one output token, that is $0.114 per million tokens for Granite-4.0-H-Small versus $0.125 for Apertus 8B (1.1× as much) and $0.15 for Devstral Small 2 (1.3× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Apertus 8B has not been scored yet, Devstral Small 2 has not been scored yet and Granite-4.0-H-Small has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Apertus 8B, Devstral Small 2 and Granite-4.0-H-Small 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?
Devstral Small 2 has the largest context window at 262,144 tokens, against 131,072 for Granite-4.0-H-Small and 65,536 for Apertus 8B. Maximum output per response: Apertus 8B up to 8,192, Devstral Small 2 up to 262,144, Granite-4.0-H-Small up to 131,072 tokens.
Which can read images, PDFs, audio or video?
Apertus 8B accepts text; Devstral Small 2 accepts text and images; Granite-4.0-H-Small accepts text. Devstral Small 2 handles the widest range of inputs.
Are any of these open source?
Yes, all three publish their weights (Apache-2.0), so you can self-host them.
Which is newer?
Devstral Small 2 is the newest, released Dec 9, 2025. Granite-4.0-H-Small came out Oct 2, 2025; Apertus 8B came out Sep 2, 2025. Knowledge cutoff: Apertus 8B Sep 2025, Devstral Small 2 Dec 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.
