INVALSI Italian LLM Benchmark
Models answer Italian reading and grammar questions from 11 school tests. For questions with several parts, the model must answer every part correctly.
The repository lists 405 questions and 618 items across six grades. The leaderboard compares 45 models.
- Made by
- CRISP, University of Milano-Bicocca
- Released
- June 2024
- Kind
- Reading comprehension
- Level
- School
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- All public
- Measures
- Capability
- Subject
- Languages and literature
- Language
- Italian
- Interaction
- Single turn
- Items from
- Exams and contests
- Refresh
- One fixed set
- Run by
- Anyone
- Strongest model tested
- o1-mini #142 on the capability index
- Last model added
- March 2025
- Scores refreshed
- November 20th, 2026
Claude 3.5 Sonnet leads with 92.2% among models released by March 2025 Claude 3.5 Sonnet leads with 92.2% among models released by March 2025
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in March 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 92.2 Claude 3.5 Sonnet
- 90.4 o1-preview
- 88.5 Claude 3 Opus
- 87.5
Mistral Large 2
- 86.1
Llama 3.1-405B
- 86.0 GPT-4 Turbo
- 83.1 Claude 3 Sonnet
- 82.9
Llama 3.2 90B
- 82.7
Llama 3.1-70B
- 81.2 google/gemini-pro-1.5
- 80.9 google/gemini-flash-1.5
- 80.9
Mixtral 8x22B
- 80.9 o1-mini
- 80.7
Gemma 2 27B
- 79.5
Qwen2-72B
- 79.0 GPT-4o mini
- 78.0 cognitivecomputations/dolphin-2.9.1-llama-3-70b
- 78.0 Claude 3 Haiku
- 76.5
Gemini 1.0 Pro
- 75.8 GPT-4o
- 75.6
Llama 3-70B
- 75.0
Command R+
- 73.8
Gemma 2 9B
- 72.1
Nemotron-4-340B-Instruct
- 68.5
Mixtral 8x7B
- 62.4
Llama 3.2 11B
- 61.9
Mistral NeMo
- 61.8 DeepMount00/Llama-3.1-8b-Ita
- 61.1 GPT-3.5 Turbo
- 59.9 DeepMount00/Llama-3-8b-Ita
- 56.5
Llama 3.1-8B
- 56.0 swap-uniba/LLaMAntino-3-ANITA-8B-Inst-DPO-ITA
- 55.5 mii-llm/maestrale-chat-v0.4-beta
- 50.6 mistralai/Mistral-7B-Instruct-v0.3
- 50.6
Phi-3-medium-128k-instruct
- 47.9
Llama 3-8B
- 44.2 mii-community/zefiro-7b-base-ITA
- 37.7 iGeniusAI/Italia-9B-Instruct-v0.1
- 35.2 DeepMount00/Llama-3-COT-ITA
- 33.5
Phi-3-mini-128k-instruct
- 30.1 ludocomito/Minerva_3B_Ties_1.0
- 25.7 sapienzanlp/modello-italia-9b
- 25.4
Llama 3.2 3B
- 15.6
Llama 3.2 1B
- 4.90 sapienzanlp/Minerva-3B-base-v1.0
- 1 Claude 3.5 Sonnet 92.2%
- 2 o1-preview 90.4%
- 3 Claude 3 Opus 88.5%
- 4
Mistral Large 2 Jul 2024 87.5%
- 5
Llama 3.1-405B 86.1%
- 6 GPT-4 Turbo Apr 2024 86%
- 7 Claude 3 Sonnet 83.1%
- 8
Llama 3.2 90B 82.9%
- 9
Llama 3.1-70B 82.7%
- 10 google/gemini-pro-1.5 81.2%
- 11 google/gemini-flash-1.5 80.9%
- 12
Mixtral 8x22B 80.9%
- 13 o1-mini 80.9%
- 14
Gemma 2 27B 80.7%
- 15
Qwen2-72B 79.5%
- 16 GPT-4o mini 79%
- 17 cognitivecomputations/dolphin-2.9.1-llama-3-70b 78%
- 18 Claude 3 Haiku 78%
- 19
Gemini 1.0 Pro 76.5%
- 20 GPT-4o Mar 2025 75.8%
- 21
Llama 3-70B 75.6%
- 22
Command R+ 75%
- 23
Gemma 2 9B 73.8%
- 24
Nemotron-4-340B-Instruct 72.1%
- 25
Mixtral 8x7B 68.5%
- 26
Llama 3.2 11B 62.4%
- 27
Mistral NeMo 61.9%
- 28 DeepMount00/Llama-3.1-8b-Ita 61.8%
- 29 GPT-3.5 Turbo Jan 2024 61.1%
- 30 DeepMount00/Llama-3-8b-Ita 59.9%
- 31
Llama 3.1-8B 56.5%
- 32 swap-uniba/LLaMAntino-3-ANITA-8B-Inst-DPO-ITA 56%
- 33 mii-llm/maestrale-chat-v0.4-beta 55.5%
- 34 mistralai/Mistral-7B-Instruct-v0.3 50.6%
- 35
Phi-3-medium-128k-instruct 50.6%
- 36
Llama 3-8B 47.9%
- 37 mii-community/zefiro-7b-base-ITA 44.2%
- 38 iGeniusAI/Italia-9B-Instruct-v0.1 37.7%
- 39 DeepMount00/Llama-3-COT-ITA 35.2%
- 40
Phi-3-mini-128k-instruct 33.5%
- 41 ludocomito/Minerva_3B_Ties_1.0 30.1%
- 42 sapienzanlp/modello-italia-9b 25.7%
- 43
Llama 3.2 3B 25.4%
- 44
Llama 3.2 1B 15.6%
- 45 sapienzanlp/Minerva-3B-base-v1.0 4.90%
Source: Official INVALSI Italian LLM Benchmark leaderboard
- 92.2 Claude 3.5 Sonnet
- 90.4 o1-preview
- 88.5 Claude 3 Opus
- 87.5
Mistral Large 2
- 86.1
Llama 3.1-405B
- 86.0 GPT-4 Turbo
- 83.1 Claude 3 Sonnet
- 82.9
Llama 3.2 90B
- 82.7
Llama 3.1-70B
- 80.9
Mixtral 8x22B
- 80.9 o1-mini
- 80.7
Gemma 2 27B
- 79.5
Qwen2-72B
- 79.0 GPT-4o mini
- 78.0 Claude 3 Haiku
- 76.5
Gemini 1.0 Pro
- 75.8 GPT-4o
- 75.6
Llama 3-70B
- 75.0
Command R+
- 73.8
Gemma 2 9B
- 72.1
Nemotron-4-340B-Instruct
- 68.5
Mixtral 8x7B
- 62.4
Llama 3.2 11B
- 61.9
Mistral NeMo
- 61.1 GPT-3.5 Turbo
- 56.5
Llama 3.1-8B
- 50.6
Phi-3-medium-128k-instruct
- 47.9
Llama 3-8B
- 33.5
Phi-3-mini-128k-instruct
- 25.4
Llama 3.2 3B
- 15.6
Llama 3.2 1B
- 1 Claude 3.5 Sonnet 92.2%
- 2 o1-preview 90.4%
- 3 Claude 3 Opus 88.5%
- 4
Mistral Large 2 Jul 2024 87.5%
- 5
Llama 3.1-405B 86.1%
- 6 GPT-4 Turbo Apr 2024 86%
- 7 Claude 3 Sonnet 83.1%
- 8
Llama 3.2 90B 82.9%
- 9
Llama 3.1-70B 82.7%
- 10
Mixtral 8x22B 80.9%
- 11 o1-mini 80.9%
- 12
Gemma 2 27B 80.7%
- 13
Qwen2-72B 79.5%
- 14 GPT-4o mini 79%
- 15 Claude 3 Haiku 78%
- 16
Gemini 1.0 Pro 76.5%
- 17 GPT-4o Mar 2025 75.8%
- 18
Llama 3-70B 75.6%
- 19
Command R+ 75%
- 20
Gemma 2 9B 73.8%
- 21
Nemotron-4-340B-Instruct 72.1%
- 22
Mixtral 8x7B 68.5%
- 23
Llama 3.2 11B 62.4%
- 24
Mistral NeMo 61.9%
- 25 GPT-3.5 Turbo Jan 2024 61.1%
- 26
Llama 3.1-8B 56.5%
- 27
Phi-3-medium-128k-instruct 50.6%
- 28
Llama 3-8B 47.9%
- 29
Phi-3-mini-128k-instruct 33.5%
- 30
Llama 3.2 3B 25.4%
- 31
Llama 3.2 1B 15.6%
Source: Official INVALSI Italian LLM Benchmark leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude 3.5 Sonnet | 92.2% | Jun 2024 | Official leaderboard | |
| 2 | o1-preview | 90.4% | Sep 2024 | Official leaderboard | |
| 3 | Claude 3 Opus | 88.5% | Feb 2024 | Official leaderboard | |
| 4 | | 87.5% | Jul 2024 | Official leaderboard | |
| 5 | | 86.1% | Jul 2024 | Official leaderboard | |
| 6 | GPT-4 Turbo Apr 2024 | 86% | Apr 2024 | Official leaderboard | |
| 7 | Claude 3 Sonnet | 83.1% | Feb 2024 | Official leaderboard | |
| 8 | | 82.9% | Sep 2024 | Official leaderboard | |
| 9 | | 82.7% | Jul 2024 | Official leaderboard | |
| 10 | google/gemini-pro-1.5 closed weights | 81.2% | Official leaderboard | ||
| 11 | google/gemini-flash-1.5 closed weights | 80.9% | Official leaderboard | ||
| 12 | | 80.9% | Apr 2024 | Official leaderboard | |
| 13 | o1-mini | 80.9% | Sep 2024 | Official leaderboard | |
| 14 | | 80.7% | Jun 2024 | Official leaderboard | |
| 15 | | 79.5% | Jun 2024 | Official leaderboard | |
| 16 | GPT-4o mini | 79% | Jul 2024 | Official leaderboard | |
| 17 | cognitivecomputations/dolphin-2.9.1-llama-3-70b open weights | 78% | Official leaderboard | ||
| 18 | Claude 3 Haiku | 78% | Mar 2024 | Official leaderboard | |
| 19 | | 76.5% | Dec 2023 | Official leaderboard | |
| 20 | GPT-4o Mar 2025 | 75.8% | Mar 2025 | Official leaderboard | |
| 21 | | 75.6% | Apr 2024 | Official leaderboard | |
| 22 | | 75% | Aug 2024 | Official leaderboard | |
| 23 | | 73.8% | Jun 2024 | Official leaderboard | |
| 24 | | 72.1% | Jun 2024 | Official leaderboard | |
| 25 | | 68.5% | Dec 2023 | Official leaderboard | |
| 26 | | 62.4% | Sep 2024 | Official leaderboard | |
| 27 | | 61.9% | Jul 2024 | Official leaderboard | |
| 28 | DeepMount00/Llama-3.1-8b-Ita open weights | 61.8% | Official leaderboard | ||
| 29 | GPT-3.5 Turbo Jan 2024 | 61.1% | Jan 2024 | Official leaderboard | |
| 30 | DeepMount00/Llama-3-8b-Ita open weights | 59.9% | Official leaderboard | ||
| 31 | | 56.5% | Jul 2024 | Official leaderboard | |
| 32 | swap-uniba/LLaMAntino-3-ANITA-8B-Inst-DPO-ITA open weights | 56% | Official leaderboard | ||
| 33 | mii-llm/maestrale-chat-v0.4-beta open weights | 55.5% | Official leaderboard | ||
| 34 | mistralai/Mistral-7B-Instruct-v0.3 open weights | 50.6% | Official leaderboard | ||
| 35 | | 50.6% | May 2024 | Official leaderboard | |
| 36 | | 47.9% | Apr 2024 | Official leaderboard | |
| 37 | mii-community/zefiro-7b-base-ITA open weights | 44.2% | Official leaderboard | ||
| 38 | iGeniusAI/Italia-9B-Instruct-v0.1 open weights | 37.7% | Official leaderboard | ||
| 39 | DeepMount00/Llama-3-COT-ITA open weights | 35.2% | Official leaderboard | ||
| 40 | | 33.5% | Apr 2024 | Official leaderboard | |
| 41 | ludocomito/Minerva_3B_Ties_1.0 open weights | 30.1% | Official leaderboard | ||
| 42 | sapienzanlp/modello-italia-9b open weights | 25.7% | Official leaderboard | ||
| 43 | | 25.4% | Sep 2024 | Official leaderboard | |
| 44 | | 15.6% | Sep 2024 | Official leaderboard | |
| 45 | sapienzanlp/Minerva-3B-base-v1.0 open weights | 4.90% | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude 3.5 Sonnet | 92.2% | Jun 2024 | Official leaderboard | |
| 2 | o1-preview | 90.4% | Sep 2024 | Official leaderboard | |
| 3 | Claude 3 Opus | 88.5% | Feb 2024 | Official leaderboard | |
| 4 | | 87.5% | Jul 2024 | Official leaderboard | |
| 5 | | 86.1% | Jul 2024 | Official leaderboard | |
| 6 | GPT-4 Turbo Apr 2024 | 86% | Apr 2024 | Official leaderboard | |
| 7 | Claude 3 Sonnet | 83.1% | Feb 2024 | Official leaderboard | |
| 8 | | 82.9% | Sep 2024 | Official leaderboard | |
| 9 | | 82.7% | Jul 2024 | Official leaderboard | |
| 10 | | 80.9% | Apr 2024 | Official leaderboard | |
| 11 | o1-mini | 80.9% | Sep 2024 | Official leaderboard | |
| 12 | | 80.7% | Jun 2024 | Official leaderboard | |
| 13 | | 79.5% | Jun 2024 | Official leaderboard | |
| 14 | GPT-4o mini | 79% | Jul 2024 | Official leaderboard | |
| 15 | Claude 3 Haiku | 78% | Mar 2024 | Official leaderboard | |
| 16 | | 76.5% | Dec 2023 | Official leaderboard | |
| 17 | GPT-4o Mar 2025 | 75.8% | Mar 2025 | Official leaderboard | |
| 18 | | 75.6% | Apr 2024 | Official leaderboard | |
| 19 | | 75% | Aug 2024 | Official leaderboard | |
| 20 | | 73.8% | Jun 2024 | Official leaderboard | |
| 21 | | 72.1% | Jun 2024 | Official leaderboard | |
| 22 | | 68.5% | Dec 2023 | Official leaderboard | |
| 23 | | 62.4% | Sep 2024 | Official leaderboard | |
| 24 | | 61.9% | Jul 2024 | Official leaderboard | |
| 25 | GPT-3.5 Turbo Jan 2024 | 61.1% | Jan 2024 | Official leaderboard | |
| 26 | | 56.5% | Jul 2024 | Official leaderboard | |
| 27 | | 50.6% | May 2024 | Official leaderboard | |
| 28 | | 47.9% | Apr 2024 | Official leaderboard | |
| 29 | | 33.5% | Apr 2024 | Official leaderboard | |
| 30 | | 25.4% | Sep 2024 | Official leaderboard | |
| 31 | | 15.6% | Sep 2024 | Official leaderboard |
Rankings by part
We read 28 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as 2 - MC, 2 - MCC and 5 - CL. Claude 3.5 Sonnet leads 9 of them, and 9 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
2 - MC
31 models we track have a score here.
- 1 Claude 3 Opus ↑2 100%
- 2 Claude 3 Sonnet ↑5 100%
- 3 Claude 3 Haiku ↑12 100%
- 4 GPT-4 Turbo Apr 2024 ↑2 100%
- 5 Claude 3.5 Sonnet ↓4 100%
- 6
Llama 3.1-405B ↓1 100%
- 7 o1-preview ↓5 100%
- 8
Mixtral 8x7B ↑14 96.9%
- 9
Gemini 1.0 Pro ↑7 96.9%
- 10
Llama 3-70B ↑8 96.9%
2 - MCC
31 models we track have a score here.
- 1 Claude 3 Opus ↑2 100%
- 2 Claude 3 Sonnet ↑5 100%
- 3 GPT-4 Turbo Apr 2024 ↑3 100%
- 4
Qwen2-72B ↑9 100%
- 5 Claude 3.5 Sonnet ↓4 100%
- 6
Llama 3.1-405B ↓1 100%
- 7
Mistral Large 2 Jul 2024 ↓3 100%
- 8 o1-mini ↑3 100%
- 9 o1-preview ↓7 100%
- 10
Llama 3.2 90B ↓2 100%
5 - CL
31 models we track have a score here.
- 1 Claude 3 Opus ↑2 100%
- 2 Claude 3 Sonnet ↑5 100%
- 3 GPT-4 Turbo Apr 2024 ↑3 100%
- 4
Qwen2-72B ↑9 100%
- 5
Nemotron-4-340B-Instruct ↑16 100%
- 6 Claude 3.5 Sonnet ↓5 100%
- 7
Llama 3.1-405B ↓2 100%
- 8
Mistral Large 2 Jul 2024 ↓4 100%
- 9
Command R+ ↑10 100%
- 10 o1-preview ↓8 100%
5 - MC
31 models we track have a score here.
- 1 Claude 3.5 Sonnet 100%
- 2 Claude 3 Opus ↑1 98.3%
- 3 Claude 3 Sonnet ↑4 96.7%
- 4
Llama 3.1-405B ↑1 96.7%
- 5
Mistral Large 2 Jul 2024 ↓1 96.7%
- 6 o1-preview ↓4 96.7%
- 7 GPT-4o mini ↑7 93.3%
- 8
Llama 3.1-70B ↑1 93.3%
- 9
Llama 3.2 90B ↓1 93.3%
- 10 Claude 3 Haiku ↑5 91.7%
5 - MCC
31 models we track have a score here.
- 1 Claude 3 Sonnet ↑6 85.7%
- 2 Claude 3.5 Sonnet ↓1 85.7%
- 3 o1-preview ↓1 85.7%
- 4 Claude 3 Opus ↓1 71.4%
- 5 GPT-4 Turbo Apr 2024 ↑1 71.4%
- 6
Nemotron-4-340B-Instruct ↑15 71.4%
- 7
Gemma 2 27B ↑5 71.4%
- 8
Llama 3.1-405B ↓3 71.4%
- 9 GPT-4o Mar 2025 ↑8 71.4%
- 10
Mixtral 8x22B 57.1%
5 - RB
31 models we track have a score here.
- 1 Claude 3 Opus ↑2 100%
- 2 Claude 3 Sonnet ↑5 100%
- 3 GPT-4 Turbo Apr 2024 ↑3 100%
- 4
Mixtral 8x22B ↑6 100%
- 5 Claude 3.5 Sonnet ↓4 100%
- 6
Mistral Large 2 Jul 2024 ↓2 100%
- 7 o1-preview ↓5 100%
- 8 GPT-4o Mar 2025 ↑9 100%
- 9
Mixtral 8x7B ↑13 0%
- 10
Gemini 1.0 Pro ↑6 0%
5 - RU
31 models we track have a score here.
- 1
Llama 3.1-405B ↑4 83.3%
- 2 o1-mini ↑9 83.3%
- 3 o1-preview ↓1 83.3%
- 4 GPT-4 Turbo Apr 2024 ↑2 66.7%
- 5
Mixtral 8x22B ↑5 66.7%
- 6
Mistral Large 2 Jul 2024 ↓2 66.7%
- 7 GPT-3.5 Turbo Jan 2024 ↑18 50%
- 8 Claude 3 Sonnet ↓1 50%
- 9
Qwen2-72B ↑4 50%
- 10 Claude 3.5 Sonnet ↓9 50%
6 - MC
31 models we track have a score here.
- 1 Claude 3.5 Sonnet 97.2%
- 2 o1-preview 95.8%
- 3
Llama 3.2 90B ↑5 95.8%
- 4
Llama 3.1-70B ↑5 94.4%
- 5 Claude 3 Opus ↓2 93%
- 6
Llama 3.1-405B ↓1 91.6%
- 7
Mistral Large 2 Jul 2024 ↓3 90.1%
- 8 Claude 3 Sonnet ↓1 88.7%
- 9
Gemma 2 27B ↑3 88.7%
- 10 GPT-4o mini ↑4 88.7%
6 - MCC
31 models we track have a score here.
- 1 GPT-4 Turbo Apr 2024 ↑5 100%
- 2
Qwen2-72B ↑11 100%
- 3 Claude 3.5 Sonnet ↓2 100%
- 4
Llama 3.1-405B ↑1 100%
- 5
Mistral Large 2 Jul 2024 ↓1 100%
- 6 o1-mini ↑5 100%
- 7 o1-preview ↓5 100%
- 8 Claude 3 Opus ↓5 85.7%
- 9
Mixtral 8x22B ↑1 85.7%
- 10 GPT-4o Mar 2025 ↑7 85.7%
6 - RU
31 models we track have a score here.
- 1 GPT-4 Turbo Apr 2024 ↑5 100%
- 2
Mistral Large 2 Jul 2024 ↑2 100%
- 3 Claude 3 Opus 88.9%
- 4
Mixtral 8x22B ↑6 88.9%
- 5 Claude 3.5 Sonnet ↓4 88.9%
- 6
Llama 3.1-70B ↑3 88.9%
- 7 o1-preview ↓5 88.9%
- 8
Llama 3.2 90B 88.9%
- 9 GPT-4o Mar 2025 ↑8 88.9%
- 10 Claude 3 Haiku ↑5 77.8%
8 - MC
31 models we track have a score here.
- 1 o1-preview ↑1 96.3%
- 2 Claude 3.5 Sonnet ↓1 95.1%
- 3 Claude 3 Opus 93.8%
- 4
Mistral Large 2 Jul 2024 93.8%
- 5 GPT-4 Turbo Apr 2024 ↑1 92.6%
- 6
Mixtral 8x22B ↑4 91.4%
- 7
Llama 3.1-405B ↓2 91.4%
- 8
Gemini 1.0 Pro ↑8 88.9%
- 9
Llama 3.2 90B ↓1 88.9%
- 10 Claude 3 Sonnet ↓3 87.6%
8 - MCC
31 models we track have a score here.
- 1 Claude 3 Haiku ↑14 100%
- 2 GPT-4 Turbo Apr 2024 ↑4 100%
- 3
Qwen2-72B ↑10 100%
- 4
Nemotron-4-340B-Instruct ↑17 100%
- 5 Claude 3.5 Sonnet ↓4 100%
- 6
Llama 3.1-405B ↓1 100%
- 7
Llama 3.1-70B ↑2 100%
- 8
Mistral Large 2 Jul 2024 ↓4 100%
- 9 o1-preview ↓7 100%
- 10
Llama 3.2 90B ↓2 100%
8 - RU
31 models we track have a score here.
- 1
Gemini 1.0 Pro ↑15 100%
- 2 Claude 3 Opus ↑1 100%
- 3 GPT-4 Turbo Apr 2024 ↑3 100%
- 4
Mixtral 8x22B ↑6 100%
- 5
Nemotron-4-340B-Instruct ↑16 100%
- 6 Claude 3.5 Sonnet ↓5 100%
- 7
Gemma 2 27B ↑5 100%
- 8
Mistral Large 2 Jul 2024 ↓4 100%
- 9
Command R+ ↑10 100%
- 10 o1-mini ↑1 100%
10 - CL
31 models we track have a score here.
- 1 GPT-3.5 Turbo Jan 2024 ↑24 50%
- 2 Claude 3 Opus ↑1 50%
- 3 Claude 3 Sonnet ↑4 50%
- 4 Claude 3 Haiku ↑11 50%
- 5 GPT-4 Turbo Apr 2024 ↑1 50%
- 6
Qwen2-72B ↑7 50%
- 7
Nemotron-4-340B-Instruct ↑14 50%
- 8 Claude 3.5 Sonnet ↓7 50%
- 9 GPT-4o mini ↑5 50%
- 10
Llama 3.1-70B ↓1 50%
10 - MC
31 models we track have a score here.
- 1 o1-preview ↑1 95.8%
- 2 Claude 3.5 Sonnet ↓1 93.8%
- 3 GPT-4 Turbo Apr 2024 ↑3 89.6%
- 4
Llama 3.1-405B ↑1 87.5%
- 5
Mistral Large 2 Jul 2024 ↓1 87.5%
- 6 Claude 3 Opus ↓3 85.4%
- 7
Gemma 2 27B ↑5 85.4%
- 8 o1-mini ↑3 85.4%
- 9 Claude 3 Sonnet ↓2 81.2%
- 10 GPT-4o mini ↑4 81.2%
10 - MCC
31 models we track have a score here.
- 1 Claude 3.5 Sonnet 69.2%
- 2
Gemma 2 27B ↑10 69.2%
- 3 GPT-4o mini ↑11 69.2%
- 4 o1-mini ↑7 69.2%
- 5 Claude 3 Opus ↓2 61.5%
- 6
Mixtral 8x22B ↑4 61.5%
- 7
Qwen2-72B ↑6 61.5%
- 8
Gemma 2 9B ↑12 61.5%
- 9
Llama 3.1-405B ↓4 61.5%
- 10
Mistral Large 2 Jul 2024 ↓6 61.5%
10 - RU
31 models we track have a score here.
- 1 o1-mini ↑10 78.6%
- 2 GPT-3.5 Turbo Jan 2024 ↑23 71.4%
- 3 Claude 3 Opus 71.4%
- 4
Mixtral 8x22B ↑6 71.4%
- 5
Llama 3-70B ↑13 71.4%
- 6
Llama 3.1-70B ↑3 71.4%
- 7
Mistral Large 2 Jul 2024 ↓3 71.4%
- 8 o1-preview ↓6 71.4%
- 9
Llama 3.2 90B ↓1 71.4%
- 10
Gemini 1.0 Pro ↑6 64.3%
13 - MC
31 models we track have a score here.
- 1 Claude 3.5 Sonnet 92.9%
- 2 Claude 3 Opus ↑1 90.5%
- 3
Mistral Large 2 Jul 2024 ↑1 83.3%
- 4 GPT-4 Turbo Apr 2024 ↑2 81%
- 5
Llama 3.1-405B 81%
- 6 Claude 3 Sonnet ↑1 78.6%
- 7
Llama 3.1-70B ↑2 78.6%
- 8 o1-preview ↓6 78.6%
- 9
Llama 3-70B ↑9 76.2%
- 10
Qwen2-72B ↑3 76.2%
13 - MCC
31 models we track have a score here.
- 1 Claude 3.5 Sonnet 62.5%
- 2 Claude 3 Opus ↑1 25%
- 3
Mixtral 8x22B ↑7 25%
- 4
Gemma 2 27B ↑8 25%
- 5 Claude 3 Sonnet ↑2 12.5%
- 6 Claude 3 Haiku ↑9 12.5%
- 7 GPT-4 Turbo Apr 2024 ↓1 12.5%
- 8
Qwen2-72B ↑5 12.5%
- 9
Nemotron-4-340B-Instruct ↑12 12.5%
- 10
Gemma 2 9B ↑10 12.5%
LI (108)
31 models we track have a score here.
- 1 Claude 3.5 Sonnet 92.6%
- 2 Claude 3 Opus ↑1 91.7%
- 3 Claude 3 Sonnet ↑4 87%
- 4
Mistral Large 2 Jul 2024 87%
- 5 GPT-4 Turbo Apr 2024 ↑1 86.1%
- 6 o1-preview ↓4 86.1%
- 7
Llama 3.1-405B ↓2 85.2%
- 8
Mixtral 8x22B ↑2 84.3%
- 9
Qwen2-72B ↑4 84.3%
- 10
Llama 3-70B ↑8 83.3%
RM (179)
31 models we track have a score here.
- 1 Claude 3.5 Sonnet 95%
- 2 o1-preview 92.7%
- 3 Claude 3 Opus 91.6%
- 4 Claude 3 Sonnet ↑3 90.5%
- 5 GPT-4 Turbo Apr 2024 ↑1 89.9%
- 6
Mistral Large 2 Jul 2024 ↓2 89.9%
- 7
Gemma 2 27B ↑5 88.8%
- 8
Llama 3.2 90B 88.8%
- 9
Llama 3.1-405B ↓4 87.7%
- 10
Llama 3.1-70B ↓1 87.2%
RC (33)
31 models we track have a score here.
- 1 o1-preview ↑1 87.9%
- 2 Claude 3.5 Sonnet ↓1 84.8%
- 3
Llama 3.1-405B ↑2 84.8%
- 4
Gemini 1.0 Pro ↑12 81.8%
- 5 GPT-4 Turbo Apr 2024 ↑1 81.8%
- 6
Mixtral 8x22B ↑4 81.8%
- 7 GPT-4o mini ↑7 81.8%
- 8
Llama 3.1-70B ↑1 81.8%
- 9
Mistral Large 2 Jul 2024 ↓5 81.8%
- 10
Command R+ ↑9 81.8%
WF (7)
31 models we track have a score here.
- 1 Claude 3 Opus ↑2 100%
- 2 Claude 3 Sonnet ↑5 100%
- 3 Claude 3.5 Sonnet ↓2 100%
- 4
Llama 3.1-405B ↑1 100%
- 5
Llama 3.1-70B ↑4 100%
- 6 o1-preview ↓4 100%
- 7
Qwen2-72B ↑6 85.7%
- 8
Gemma 2 27B ↑4 85.7%
- 9 GPT-4o mini ↑5 85.7%
- 10
Mistral Large 2 Jul 2024 ↓6 85.7%
LS (29)
31 models we track have a score here.
- 1 Claude 3.5 Sonnet 93.1%
- 2
Mistral Large 2 Jul 2024 ↑2 93.1%
- 3 o1-preview ↓1 93.1%
- 4 GPT-4 Turbo Apr 2024 ↑2 86.2%
- 5 o1-mini ↑6 86.2%
- 6 Claude 3 Opus ↓3 82.8%
- 7
Llama 3.1-405B ↓2 82.8%
- 8
Qwen2-72B ↑5 79.3%
- 9
Llama 3.1-70B 79.3%
- 10
Llama 3.2 90B ↓2 79.3%
MO (24)
31 models we track have a score here.
- 1 o1-preview ↑1 95.8%
- 2 Claude 3.5 Sonnet ↓1 87.5%
- 3 o1-mini ↑8 87.5%
- 4
Mixtral 8x22B ↑6 83.3%
- 5
Llama 3.1-405B 83.3%
- 6
Mistral Large 2 Jul 2024 ↓2 83.3%
- 7 GPT-4 Turbo Apr 2024 ↓1 79.2%
- 8 GPT-4o Mar 2025 ↑9 79.2%
- 9 Claude 3 Opus ↓6 75%
- 10 Claude 3 Sonnet ↓3 75%
SP (4)
31 models we track have a score here.
- 1 Claude 3 Opus ↑2 50%
- 2 GPT-4 Turbo Apr 2024 ↑4 50%
- 3
Llama 3.1-405B ↑2 50%
- 4 o1-mini ↑7 50%
- 5 o1-preview ↓3 50%
- 6
Nemotron-4-340B-Instruct ↑15 25%
- 7 Claude 3.5 Sonnet ↓6 25%
- 8
Llama 3.1-70B ↑1 25%
- 9
Mistral Large 2 Jul 2024 ↓5 25%
- 10
Llama 3.2 90B ↓2 25%
SY (19)
31 models we track have a score here.
- 1 Claude 3.5 Sonnet 94.7%
- 2 GPT-4o Mar 2025 ↑15 94.7%
- 3 Claude 3 Opus 89.5%
- 4 o1-preview ↓2 89.5%
- 5
Llama 3.1-405B 84.2%
- 6
Mistral Large 2 Jul 2024 ↓2 84.2%
- 7 o1-mini ↑4 84.2%
- 8
Qwen2-72B ↑5 79%
- 9
Llama 3.1-70B 79%
- 10 GPT-4 Turbo Apr 2024 ↓4 73.7%
TP (6)
31 models we track have a score here.
- 1 Claude 3 Sonnet ↑6 100%
- 2 GPT-4 Turbo Apr 2024 ↑4 100%
- 3
Qwen2-72B ↑10 100%
- 4
Llama 3.1-405B ↑1 100%
- 5
Mistral Large 2 Jul 2024 ↓1 100%
- 6 o1-preview ↓4 100%
- 7
Llama 3.2 90B ↑1 100%
- 8 GPT-4o Mar 2025 ↑9 100%
- 9 GPT-3.5 Turbo Jan 2024 ↑16 83.3%
- 10 Claude 3 Opus ↓7 83.3%
2 models have held number one on the official leaderboard since its release in June 2024
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official INVALSI Italian LLM Benchmark leaderboard
Claude 3 Opus on the official leaderboard passed 85% in February 2024
Models released before it came out already scored that high. All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official INVALSI Italian LLM Benchmark leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Knowledge
- MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects.
Intern-S2-397B 89.8%
- MMLU In use 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy.
Seed1.8 92.3%
- SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy.
Qwen 3.6 Max (Preview) 73.9%
- WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1.
Tencent Hy4 preview 83.9%
- BBEH micro average In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is the share of all examples answered correctly.
Gemma 4 31B IT 74.4%
- HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. Claude Fable 5.2 77.9%
Data audit
All 8 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
- Pass Independent runs Scores from the official leaderboard (45 entries).
- Pass Models scored 31 models we track, plus 14 other entries on its leaderboard.
- Note Newest score The newest came 20 months ago, in March 2025, for GPT-4o. That’s expected for a saturated test.
- Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is o1-mini (#142).
- Note Headroom Saturated. The best is 92.2%, past 85% since February 2024.
Leaderboard import 3 pass
- Pass Importer We read the board on huggingface.co: 31 rows for models we track and 14 other entries at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.