SmolLM3 family
Compact multilingual instruction model with hybrid reasoning and long-context support; the released family centers on 3B checkpoints.
Score calculation
(2 + 5 + 8 + 1 + 7) ÷ 5 = 4.6
5 of six dimensions rated. Unrated dimensions are excluded from the average.
Full scoring method ↗Setup details
HuggingFaceTB SmolLM3 3B base and instruction checkpoints.
What fits your needs
HuggingFaceTB SmolLM3 3B base and instruction checkpoints.
Good fit
Compact multilingual instruction model with hybrid reasoning and long-context support; the released family centers on 3B checkpoints.
Tradeoffs to expect
Tradeoffs
Apache 2.0 Apache 2.0 weights, training configuration, data mixture, and intermediate checkpoints are unusually inspectable at this size. Some later-stage dataset details are still incomplete, and training data permission remains a separate question.
How to switch
Local setup
The published 3B checkpoint and Transformers instructions support local use. Actual memory depends on precision, context length, and runtime; choose a supported quantization only when its source is clear.
Practical questions
Weights and license
Fetch the exact checkpoint revision and read its own license; sibling and distilled variants may inherit different terms.
Hardware
The published 3B checkpoint and Transformers instructions support local use. Actual memory depends on precision, context length, and runtime; choose a supported quantization only when its source is clear.
Training data
Treat the weight license and public web/data availability separately from evidence of permission for the training data.
This is a practices and disclosure score for the named release, not a request-level energy or emissions estimate.
See the evidence
The official card reports 384 H100 GPUs for training and a staged token count. It gives no energy, emissions, or water measurements. The rubric rewards documented reporting and controls; it does not infer high consumption from missing measurements.
1 + 1 reporting + 0 electricity + 0 controls = 2/10. This grades documented practices and transparency.
No comparable per-request electricity, emissions, water, regional serving mix, or independently reviewed model footprint is established in the cited sources.
- Reporting1/3
- The official card reports 384 H100 GPUs for training and a staged token count. It gives no energy, emissions, or water measurements.
- Electricity sourcing0/3
- No model-specific clean-electricity sourcing or hourly power evidence is published in the reviewed sources; zero means unestablished, not dirty electricity.
- Resource controls0/3
- No model-specific documented control or measured physical resource reduction is established by the reviewed release materials. Public weights or smaller variants alone do not prove reduced resource use.
Water disclosure
Water reporting has not been assessed for this choice.
The release provides some training-process information, but not a complete permission ledger for inherited and later training data.
See the evidence
The card links a public data mixture, pretraining and post-training stages, configs, and intermediate checkpoints. Named sources improve traceability, but public availability and a weight license don’t confirm permission across the training data. A public weight license governs use of those weights; it is not proof that every training source was permitted.
1 base + 3 source transparency + 1 documented permission + 0 creator controls = 5/10. Weight licensing is not treated as training data permission.
Complete source lists, source-specific rights, inherited-base provenance, synthetic-data generation permissions, and creator removal or retraining remedies remain unestablished.
- Source transparency3/3
- The card links a public data mixture, pretraining and post-training stages, configs, and intermediate checkpoints. Named sources improve traceability, but public availability and a weight license don’t confirm permission across the training data.
- Documented permission1/4
- Some named or permissively licensed material is described, but the sources don’t settle permission for the full inherited and later-stage training data.
- Creator controls0/2
- The reviewed model documentation doesn’t show creator permission covering all training data, opt-out, removal, or retraining remedy for this model family.
Privacy depends on where these released weights are run; the model family has no single prompt-retention or training route.
See the evidence
The official release makes weights available for separate deployments, but it does not define one inference host or retention policy for every user. Local and hosted operators may handle prompts differently, so a single numerical privacy grade would misstate this model-level entry.
The user-selected host, region, telemetry, logging, extensions, and provider training terms are outside the checkpoint release and require a route-specific review.
Apache 2.0 weights, training configuration, data mixture, and intermediate checkpoints are unusually inspectable at this size. Some later-stage dataset details are still incomplete, and training data permission remains a separate question.
See the evidence
Apache 2.0 weights, training configuration, data mixture, and intermediate checkpoints are unusually inspectable at this size. Some later-stage dataset details are still incomplete, and training data permission remains a separate question. This openness rating concerns artifacts and reproducibility only; it does not assess whether the original training data was authorized.
Apache 2.0 Model weights and documentation are available, but limitations described above reduce source-level reproducibility.
The checked release and license apply to the named checkpoints; sibling variants, future versions, and the underlying training data may have different terms or disclosures.
Hugging Face’s public conduct code covers community participation, not employee or model-workforce conditions.
See the evidence
Hugging Face’s Contributor Covenant addresses harassment in its online community and provides a reporting route for community misconduct. It does not set employee or contractor pay, working conditions, or supply-chain standards, so it does not earn workforce-protection points for SmolLM3 labor.
1 base + 0 conditions + 0 worker voice/remedy + 0 labor assurance = 1/10. The public community code is outside the worker-protection scope.
No Hugging Face employee conditions, annotator/evaluator terms, supplier coverage, labor audits, or worker-level grievance outcomes were established in the reviewed official sources.
- Pay & conditions0/3
- The reviewed public code governs community conduct rather than employees’ or contractors’ employment conditions.
- Worker voice & remedy0/3
- Its complaint channel addresses community misconduct; no employee or labor grievance route was established.
- Evidence & follow-through0/3
- No independent worker-level labor audit or follow-up outcome was established in the reviewed sources.
No military relationship or release-specific safeguards are established for Hugging Face in the reviewed model sources.
See the evidence
The official model card and release materials identify the model and license but do not provide a military customer, contract, deployment, or model-specific enforcement record. A relationship may still exist even if this review found no record of it.
7 base + 0 safeguard adjustment = 7/10 (0 of 9 safeguard points; every 3 adds 1). No military relationship was established in the cited sources. This ranks above known defense ties without claiming a verified clean record.
Undisclosed contracts, customer-specific deployment, end-use restrictions, and independent compliance outcomes are not established by this model-family review.
- Use restrictions0/4
- The cited model license and release documents don’t show a model-specific binding ban on military or intelligence use. This is limited to this source set.
- Enforcement & oversight0/3
- No military-specific monitoring, enforcement, or remedy mechanism is established in the reviewed model materials.
- Relationship disclosure0/2
- The model materials do not disclose a model-specific military relationship or military-use record. A relationship may still exist even if this review found no record of it.
The reviewed model-release sources don’t confirm a defense relationship; absence of disclosure is not a clean-record finding.
Military rubric ↗Hugging Face is not named in the two official BNC campaign lists checked. This is a limited list check, not BDS approval.
See the evidence
The official BNC consumer boycott/pressure guide and No Tech campaign page do not name Hugging Face among their current targets. BDS uses selective campaigns, so absence from these lists is neutral and should not be read as a clean record or approval.
This check covers only named targets in the two cited BNC campaigns; it does not cover all BDS campaigns, local groups, investments, suppliers, or downstream hosts. Lists and ownership may change.