Mistral Fine-Tuning API
Mistral AIMistral's managed fine-tuning service on its Studio platform, offering LoRA-based supervised fine-tuning for text and vision on Mistral Small and Medium, plus a Classifier Factory for classification tasks. Mistral's documentation now files the service under deprecated, legacy features. A separate SDK, mistral-finetune, covers self-hosted training.
The deprecated label is a docs classification, not evidence the API is retired: pricing still bills training and storage for the Classifier API tier. Verified 2026-08-09 via Mistral's deprecated fine-tuning guide, the Studio product page, and the API pricing page.
Openness
1 high confidence- source
- closed
- training-pipeline
- closed
- runs-on
- La-Plateforme-only
- license
- Proprietary(managed service, per-token + storage pricing)
- base-weights
- Mistral-models(open-weight variants exist)
The hosted FT API on Mistral's Studio platform (formerly branded La Plateforme) is a proprietary managed service (LoRA adapters under the hood); no source for the service. Mistral separately ships an OSS mistral-finetune SDK for self-hosting, but that is a distinct product from this managed API. Mistral's own docs file the service under 'Deprecated features > Fine-tuning API (legacy)' with a banner reading 'This feature is deprecated and is no longer actively supported', and the current API pricing catalog lists ongoing training/storage billing only for the Classifier API tier, with no separate catalog entry for general text/vision SFT. Openness is unaffected either way: still a closed, proprietary managed offering with no self-hostable implementation.
- https://mistral.ai/news/customization recorded 2026-08-09
States it will 'leverage our proprietary fine-tuning techniques with our managed fine-tuning services' for the serverless offering, and separately introduces the open source mistral-finetune SDK as the self-hosted alternative; no repository is offered for the managed service itself.
- https://docs.mistral.ai/resources/deprecated/finetuning recorded 2026-08-09
Nav files this page under 'Deprecated features' as 'Fine-tuning API (legacy)'; the page banner reads 'This feature is deprecated and is no longer actively supported.' Describes the service as 'Text and vision general fine-tuning via SFT' plus 'Classifier Factory', billed with 'a minimum fee per fine-tuning job of $4' and a monthly storage fee; no repository or self-hosted build path is offered for this managed feature.
- https://mistral.ai/pricing/api/ recorded 2026-08-09
Its 'Browse our APIs' catalog lists 'Classifier API model 3B' and 'Classifier API model 8B' fine-tuning, each billed 'minimum fee per fine-tuning job of $4' plus a per-token training cost and a monthly per-model storage cost; this is Mistral's own commercial price list, not a license grant, and no separate catalog entry prices general text/vision SFT.
- https://legal.mistral.ai/terms recorded 2026-08-09
Indexes Mistral's legal documents, listing a 'Commercial Terms of Service' for commercial/API customers and a 'License Notice' for everyone; no open source license is offered for the hosted platform itself, consistent with a proprietary managed service.
Adoption
not assessedNo standalone usage figure exists for the Mistral fine-tuning API: no jobs run, no customers tuning, no models fine-tuned, no developer count. The platform around it is widely used by developers, but nothing published is specific to fine-tuning, so no level is assigned. The absence is itself the finding for a feature sold inside a larger platform.
- https://mistral.ai/news/customization recorded 2026-08-13
managed FT feature on La Plateforme; no usage figures disclosed
Capability
3 medium confidenceA narrower managed method set than the services offering supervised, preference and reinforcement fine-tuning together: here it is LoRA-adapter supervised fine-tuning for text and vision, plus the Classifier Factory for classification models, over Mistral's own Small and Medium tier models, with no exposed full fine-tuning, DPO or RFT and no scale or precision controls. It is a real, capable managed LoRA service, but it sits below the preference- and RL-capable offerings from OpenAI, Azure, Together and Fireworks, and far below the open-source libraries that define scale. Mistral's own docs still describe both methods in a fine-tuning guide now filed under deprecated features, while the live API pricing catalog lists ongoing billing only for the Classifier tier, with no separate entry for general text or vision fine-tuning.
- https://docs.mistral.ai/resources/deprecated/finetuning recorded 2026-08-09
Confirms the recorded method set verbatim: 'Text and vision general fine-tuning via SFT' and 'Classifier Factory', with no full fine-tune, DPO, or RFT option mentioned; filed under 'Deprecated features > Fine-tuning API (legacy)' with a 'no longer actively supported' banner.
- https://mistral.ai/pricing/api/ recorded 2026-08-09
The current API catalog itemizes fine-tuning pricing only for 'Classifier API model 3B' and 'Classifier API model 8B'; no catalog card prices general text/vision SFT of Mistral Small or Medium, though the deprecated guide still documents it as a service.
Verified 2026-08-09