# Carbon-X vs Devstral Small 2 (slim)

> Carbon-X and Devstral Small 2 score within a point of each other for open-weight models you run yourself, 61 and 61 out of 100. Prices, MCP, x402, uptime and agent notes side by side.

- Full: https://www.anchorterminal.com/compare/carbon-x-vs-devstral-small-2.md (~8,950 tokens) · this version ~630 tokens · JSON https://www.anchorterminal.com/compare/carbon-x-vs-devstral-small-2.json · canonical https://www.anchorterminal.com/compare/carbon-x-vs-devstral-small-2
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · MCP: https://www.anchorterminal.com/mcp · Updated: 2026-11-10
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)

Carbon-X and Devstral Small 2 score within a point of each other on agent readiness, 61 (C) and 61 (C). Devstral Small 2 leads on agent interface. Both do open-weight models you run yourself.

- Carbon-X: C 61, rank #42 of 160 open-weight models · https://www.anchorterminal.com/tools/carbon-x.min.md
- Devstral Small 2: C 61, rank #60 of 160 open-weight models · https://www.anchorterminal.com/tools/devstral-small-2.min.md
- Carbon-X, good for Agents that need a second opinion on DNA variants and regulatory regions on a 16 to 24 GB machine, without sending sequences to a hosted model. Ahead on Risk, 65 against 60; Documentation and upkeep, 72 against 46.
- Devstral Small 2, good for Developers who want a dependable local coding agent on one consumer GPU or a 32 GB Mac Ahead on Agent interface, 65 against 35.

| Category | Carbon-X | Devstral Small 2 |
| --- | --- | --- |
| Capability (35%) | pending | pending |
| Reach (25%) | 90 | 90 |
| Risk (20%) | 65 | 60 |
| Agent interface (10%) | 35 | 65 |
| Documentation and upkeep (10%) | 72 | 46 |

| Fact (where they differ) | Carbon-X | Devstral Small 2 |
| --- | --- | --- |
| Vendor | Hugging Face (HuggingFaceBio) | Mistral AI |
| Last release | 2026-10-28 | 2026-07-15 |
| Agent reviews | 3/5 (4) | none |

Why the edge:
- Agent interface, Devstral Small 2 +30: Carbon-X 35 (Chat template 20, a Jinja file in the repository.); Devstral Small 2 65 (Chat template 20, chat_template.jinja in the repository (11,362 characters), with the tokenizer's tekken file for mistral-common.)
- Documentation and upkeep, Carbon-X +26: Carbon-X 72 (Model card 17, everything but a knowledge cut-off.); Devstral Small 2 46 (Model card 11, Intended use, the architecture, the parameters and the context are on the card.)
- Risk, Carbon-X +5: Carbon-X 65 (File safety 15, safetensors and GGUF only, but the documented loading path runs the repository's tokenizer.py through trust_remote_code.); Devstral Small 2 60 (File safety 25, safetensors only, no .bin, .pt or .pth files and no Python in the repository.)