Swiss AI InitiativeLanguage model
Apertus 70B
Apertus 70B has no overall index value yet because there are too few independent benchmark results.
PopularityRank 361 of 785
37out of 100-1 vs. last month
based on search, usage and attention
Summary
Apertus 70B is the large instruction variant in Swiss AI’s fully open language-model family. Its official card describes 15 trillion pretraining tokens, 1,811 natively supported languages, and open publication of weights, data reconstruction, and training details. A decoder-only architecture, SFT, and QRPO connect the transparent research base to an instruction-tuned release. The context is 65,536 tokens.
Swiss AI trained Apertus 70B from scratch with the AdEMAMix optimizer and xIELU activations. Apertus 8B is the smaller family partner with the same language coverage but its own weight checkpoint.
All benchmark results
Every published result with its source, test setup and comparison models.
Show all 86 published benchmark results
Page 1 of 8
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
CritPt v2: Overall (Show measurement, test conditions, and source)
- Source value
- 0
- Score
- 0
- Metric
- pass@1 accuracy
- Unit
- %
- Benchmark version
- v2
- Category
- reasoning
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- disputed
- Retrieved at
- 2026-11-02
- Methodology
- Exact AA configuration: Apertus 70B Instruct (release apertus-70b-instruct, row apertus-70b-instruct, UUID 49e70a38-4ac1-4659-b490-09b2c7ff21d6, reasoning=false, effort=not specified, suffix=not specified). AA marks this model variant as non-reasoning and reports 0 reasoning tokens. The full CritPt v2 run covers 70 questions with five repeats. The official grading server reports pass@1.. Source model ID: artificial-analysis/49e70a38-4ac1-4659-b490-09b2c7ff21d6/apertus-70b-instruct/apertus-70b-instruct. Source model name: Apertus 70B. Provider reported by source: Swiss AI Initiative. Model snapshot: Apertus 70B Instruct. Evaluator: Artificial Analysis. Dataset type: artificial-analysis-evaluation. Tool policy: Python functions, symbolic expressions, and numerical answers; official grading server. Attempts: 70 questions, five repeats per question, pass@1. Reasoning: non-reasoning. Reasoning effort: none. Score display: 0.0%. Source ID: artificial-analysis. Evidence capture ID: aa-model-page-2026-11-02-67d46be20a5b. Source payload hash: 67d46be20a5b412db6cfb60ad190441fcaa5513500b8fd47d033168d0e97b6b2. Parser version: artificial-analysis-rsc-models-v3. Imported at: 2026-11-02. Source cohort: artificial-analysis-critpt-v2:v2:standalone:non-reasoning:effort=unspecified:suffix=unspecified
- Comparison methodology
- Exact AA configuration: Apertus 70B Instruct (release apertus-70b-instruct, row apertus-70b-instruct, UUID 49e70a38-4ac1-4659-b490-09b2c7ff21d6, reasoning=false, effort=not specified, suffix=not specified). AA marks this model variant as non-reasoning and reports 0 reasoning tokens. The full CritPt v2 run covers 70 questions with five repeats. The official grading server reports pass@1.. Tool policy: Python functions, symbolic expressions, and numerical answers; official grading server. Attempts: 70 questions, five repeats per question, pass@1. Reasoning: non-reasoning. Reasoning effort: none
- Context
- This AA source field contains the original CritPt score. The earlier CritPt v2 attribution was incorrect; the value and former label remain as disputed historical evidence.. The source dataset marks this result as disputed. It remains visible as documented evidence, but is not reliable comparative proof.
IFBench: Overall (Show measurement, test conditions, and source)
- Source value
- 25.8503401360544
- Score
- 25.8503401360544
- Metric
- IFBench prompt-level loose accuracy (pass@1)
- Unit
- %
- Benchmark version
- AA test set: 294 questions
- Category
- general
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- superseded
- Retrieved at
- 2026-11-02
- Methodology
- Exact AA configuration: Apertus 70B Instruct (release apertus-70b-instruct, row apertus-70b-instruct, UUID 49e70a38-4ac1-4659-b490-09b2c7ff21d6, reasoning=false, effort=not specified, suffix=not specified). AA marks this model variant as non-reasoning and reports 0 reasoning tokens. Artificial Analysis runs 294 questions from the IFBench test set with five repeats per question. The official evaluator scores prompt-level accuracy in loose mode as pass@1.. Source model ID: artificial-analysis/49e70a38-4ac1-4659-b490-09b2c7ff21d6/apertus-70b-instruct/apertus-70b-instruct. Source model name: Apertus 70B. Provider reported by source: Swiss AI Initiative. Model snapshot: Apertus 70B Instruct. Evaluator: Artificial Analysis. Dataset type: artificial-analysis-evaluation. Tool policy: Single-turn answer evaluated with the official IFBench source code. Attempts: pass@1, 5 repeats per question. Reasoning: non-reasoning. Reasoning effort: none. Score display: 25.9%. Source ID: artificial-analysis. Evidence capture ID: aa-model-page-2026-11-02-67d46be20a5b. Source payload hash: 67d46be20a5b412db6cfb60ad190441fcaa5513500b8fd47d033168d0e97b6b2. Parser version: artificial-analysis-rsc-models-v3. Imported at: 2026-11-02. Source cohort: artificial-analysis-ifbench:AA test set: 294 questions:standalone:non-reasoning:effort=unspecified:suffix=unspecified
- Comparison methodology
- Exact AA configuration: Apertus 70B Instruct (release apertus-70b-instruct, row apertus-70b-instruct, UUID 49e70a38-4ac1-4659-b490-09b2c7ff21d6, reasoning=false, effort=not specified, suffix=not specified). AA marks this model variant as non-reasoning and reports 0 reasoning tokens. Artificial Analysis runs 294 questions from the IFBench test set with five repeats per question. The official evaluator scores prompt-level accuracy in loose mode as pass@1.. Tool policy: Single-turn answer evaluated with the official IFBench source code. Attempts: pass@1, 5 repeats per question. Reasoning: non-reasoning. Reasoning effort: none
- Context
- The often cited 58 refers to verifiable out-of-domain constraint types, not the number of questions in AA's test run.. This result is retained as historical evidence and is not interchangeable with current results.
IFBench: Overall (Show measurement, test conditions, and source)
- Source value
- 25.8503401360544
- Score
- 25.8503401360544
- Metric
- IFBench prompt-level loose accuracy (pass@1)
- Unit
- %
- Benchmark version
- AA test set: 294 questions
- Category
- general
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- active
- Retrieved at
- 2026-11-07
- Methodology
- Exact AA configuration: Apertus 70B Instruct (release apertus-70b-instruct, row apertus-70b-instruct, UUID 49e70a38-4ac1-4659-b490-09b2c7ff21d6, reasoning=false, effort=not specified, suffix=not specified). This source capture does not provide an average reasoning-token count. Artificial Analysis runs 294 questions from the IFBench test set with five repeats per question. The official evaluator scores prompt-level accuracy in loose mode as pass@1.. Source model ID: artificial-analysis/49e70a38-4ac1-4659-b490-09b2c7ff21d6/apertus-70b-instruct/apertus-70b-instruct. Source model name: Apertus 70B. Provider reported by source: Swiss AI Initiative. Model snapshot: Apertus 70B Instruct. Evaluator: Artificial Analysis. Dataset type: artificial-analysis-evaluation. Tool policy: Single-turn answer evaluated with the official IFBench source code. Attempts: pass@1, 5 repeats per question. Reasoning: non-reasoning. Reasoning effort: none. Score display: 25.9%. Source ID: artificial-analysis. Evidence capture ID: aa-model-page-2026-11-07-b96e3ffc8c5b. Source payload hash: b96e3ffc8c5b052aa56f96d05e89d02b8cc234a75085820fca0ca2a39e498274. Parser version: artificial-analysis-rsc-models-v4. Imported at: 2026-11-07. Source cohort: artificial-analysis-ifbench:AA test set: 294 questions:standalone:non-reasoning:effort=unspecified:suffix=unspecified
- Comparison methodology
- Exact AA configuration: Apertus 70B Instruct (release apertus-70b-instruct, row apertus-70b-instruct, UUID 49e70a38-4ac1-4659-b490-09b2c7ff21d6, reasoning=false, effort=not specified, suffix=not specified). This source capture does not provide an average reasoning-token count. Artificial Analysis runs 294 questions from the IFBench test set with five repeats per question. The official evaluator scores prompt-level accuracy in loose mode as pass@1.. Tool policy: Single-turn answer evaluated with the official IFBench source code. Attempts: pass@1, 5 repeats per question. Reasoning: non-reasoning. Reasoning effort: none
- Context
- The often cited 58 refers to verifiable out-of-domain constraint types, not the number of questions in AA's test run.
MMLU (Show measurement, test conditions, and source)
- Source value
- 69.6
- Score
- 69.6
- Metric
- Exact Match
- Unit
- %
- Category
- knowledge
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Apertus v1 Technical Report v2, Table 17
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Exact Apertus-70B-Instruct row. Table 22: mmlu_flan_cot_zeroshot, MCQ generation, CoT, 0-shot, chat template, no multi-turn, one language.
Global-MMLU (Show measurement, test conditions, and source)
- Source value
- 62.7
- Score
- 62.7
- Metric
- Exact Match
- Unit
- %
- Category
- multilingual
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Apertus v1 Technical Report v2, Table 17
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Exact Apertus-70B-Instruct row. Table 22: global_mmlu_gen_0shot, MCQ generation, no CoT, 0-shot, chat template, no multi-turn, 15 languages.
TruthfulQA (Show measurement, test conditions, and source)
- Source value
- 61.2
- Score
- 61.2
- Metric
- Accuracy
- Unit
- %
- Category
- knowledge
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Apertus v1 Technical Report v2, Table 17
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Exact Apertus-70B-Instruct row. Table 22: truthfulqa_mc2, MCQ likelihood, no CoT, 6-shot, chat template, no multi-turn, one language.
TruthfulQA Multilingual (Show measurement, test conditions, and source)
- Source value
- 53.7
- Score
- 53.7
- Metric
- Accuracy
- Unit
- %
- Category
- multilingual
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Apertus v1 Technical Report v2, Table 17
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Exact Apertus-70B-Instruct row. Table 22: truthfulqa_multilingual_mc2, MCQ likelihood, no CoT, 5-shot, chat template, no multi-turn, 31 languages.
HellaSwag (Show measurement, test conditions, and source)
- Source value
- 78.1
- Score
- 78.1
- Metric
- Normalized Accuracy
- Unit
- %
- Category
- knowledge
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Apertus v1 Technical Report v2, Table 17
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Exact Apertus-70B-Instruct row. Table 22: hellaswag, MCQ likelihood, no CoT, 0-shot, chat template, no multi-turn, one language.
HellaSwag Multilingual (Show measurement, test conditions, and source)
- Source value
- 55.3
- Score
- 55.3
- Metric
- Normalized Accuracy
- Unit
- %
- Category
- multilingual
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Apertus v1 Technical Report v2, Table 17
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Exact Apertus-70B-Instruct row. Table 22: hellaswag_multilingual, MCQ likelihood, no CoT, 0-shot, chat template, no multi-turn, 31 languages.
BBH (Show measurement, test conditions, and source)
- Source value
- 64.2
- Score
- 64.2
- Metric
- Exact Match
- Unit
- %
- Category
- reasoning
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Apertus v1 Technical Report v2, Table 19
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Exact Apertus-70B-Instruct row. Table 22: bbh, generation, CoT, 3-shot, chat template and multi-turn demonstrations, one language.
DROP (Show measurement, test conditions, and source)
- Source value
- 50.8
- Score
- 50.8
- Metric
- F1
- Unit
- %
- Category
- reasoning
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Apertus v1 Technical Report v2, Table 19
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Exact Apertus-70B-Instruct row. Table 22: drop, generation, no CoT, 3-shot, chat template and multi-turn demonstrations, one language.
ACP-Bench Bool (Show measurement, test conditions, and source)
- Source value
- 62.9
- Score
- 62.9
- Metric
- Exact Match
- Unit
- %
- Category
- reasoning
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Apertus v1 Technical Report v2, Table 19
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Exact Apertus-70B-Instruct row. Table 22: acp_bench_bool, MCQ generation, CoT, 2-shot, chat template and multi-turn demonstrations, one language.
Page 1 of 8
Popularity compared
How often the model is searched for, used and mentioned, compared with every model of its category.
Popularity index
The 100 most popular models of this category, 0 to 100, higher is more popular
20 of 101 models
As of November 2026 · Source: gradually.ai popularity index
Over time
Weekly values from October 18, 2026 to November 7, 2026
Apertus 70BLlama 3.3 Nemotron Super 49B v1.5 37GLM-4.6 37Index-Translate-9B 37Reflection Beam 37
Method changes in the timeline, a jump at these points is no market event:
- 1November 1, 2026: Value only with search or usage data, 30-day attention
What is gaining popularity right now, across models, vendors, harnesses and AI subscriptions, is collected in All AI trends.
Similar models
More models of the same kind as Apertus 70B, chosen by model line, developer, size, price and release date.
- Apertus 8BSwiss AI Initiative · same developer
- Hermes 4 70BNous Research · similar size
- Qwen 3 Next 80B A3B InstructAlibaba · similar size
- Qwen 3 Next 80B A3B ThinkingAlibaba · similar size
- Llama 3.1 Nemotron 70B InstructNVIDIA · similar size
- DeepSeek-R1-Distill-Llama-70BDeepSeek · similar size
- Llama-3.3-70B-InstructMeta · similar size
- Llama 3.3 70BMeta · similar size
All specifications
Architecture
Input and output
Watermarking and provenance
AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.
- AI watermarking
- Unknown
- no source
- Watermarking technique
- Not known
- no source
Licence and access
Frequently asked questions about Apertus 70B
What is the knowledge cutoff of Apertus 70B?
September 2025, according to Swiss AI Initiative. That is the month of release.
Stated by Swiss AI Initiative
SepKnowledge cutoffSeptember 2025ReleasedSep 2, 2025
How large is the context window of Apertus 70B?
65,536 tokens. That is how much text Apertus 70B reads at once, about 100 A4 pages.
Page count as a rule of thumb from Artificial Analysis
GLM-5.1200K
DeepSeek-V3 0324164K
GLM-4.5131K
GPT OSS 120B131K
DeepSeek-V3.2128K
DeepSeek-R1128K
Apertus 70BRank 103 of 10366K
Can I run Apertus 70B locally?
Yes, the weights are open.
Stated by Swiss AI Initiative
- Open weights
- yes
- License
- Apache 2.0
- Parameters
- 70B
- active per token
- n/a
Sources and data date
Show all 13 sources
Benchmark sources at a glance
Which independent evaluators tested this model. This is coverage, not a quality ranking.
| Evaluator | Test series | Composite indices | Recorded active measurements |
|---|---|---|---|
| Artificial Analysis | 1 | 0 | 1 |
Without vendor claims and historical or disputed results; those stay labelled in the single values.
| Type | Evidence and data date |
|---|---|
| Input | Documented modalities. Retrieved October 27, 2026. |
| Output | Documented modalities. Retrieved October 27, 2026. |
| Research status | Research date October 15, 2026. 3 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication. |
| Not published | evaluation-configuration Note The card lists exact model rows and tasks but not score units, prompts, few-shot settings, training/evaluation mode, or run dates. |
| Not found in the inspected sources | excluded-checkpoint-observations Note The six rows match Table 14 of the Apertus v1 technical report's pretraining evaluation. They describe pretrained Apertus-8B/70B checkpoints, not the separately post-trained Instruct profiles. |
| Not published | sample-count-and-confidence-intervals Original wording Die Tabellen veröffentlichen für diese Scorezellen keine ausgewerteten Fallzahlen oder Konfidenzintervalle. Korpusgrößen, Shot-Angaben und Sprachzahlen werden nicht als Ergebnis-N umgedeutet. |
| Not published | exact-decoding-and-run-provenance Original wording Tabelle 22 nennt für die meisten Evaluationen Benchmark-Metrik, Tasktyp, CoT, Shots, Chat-Template, Mehrturn- und Sprachumfang, aber keine vollständigen Harness-/Prompt-/Decoding-Versionen oder konkreten Lauftermine. Tabellen 18, 23 und 26 nennen außerdem keine Score-Einheit. |
| Additional source | Hugging Face (retrieved October 15, 2026; October 17, 2026) · Apertus 70B model card · Editorial description reviewed October 4, 2026 |
| Additional source | Artificial Analysis (retrieved November 2, 2026) |
| Additional source | Artificial Analysis (retrieved November 2, 2026; November 7, 2026) |
| Additional source | Apertus v1 Technical Report v2, Table 17 (retrieved October 17, 2026) |
| Additional source | LLMs replicate metaphor norms based on word co-occurrence but struggle with topic-vehicle mappings (retrieved October 19, 2026) |
| Additional source | Apertus technical report · Editorial description reviewed October 4, 2026 |
| Additional source | OpenRouter: Popularity index (derived), CC BY 4.0 |
| Additional source | Vercel AI Gateway: Popularity index (derived), CC BY 4.0 |
| Additional source | Ollama: cumulative downloads for the popularity index |
| Additional source | Replicate: cumulative runs for the popularity index |
| Additional source | Google Trending Now: limited feed selection for Germany and the US |
| Additional source | Creative Commons Attribution 4.0 |