Benchmark Almanac

Which AI Model Knows Israel Best?

Grok 4.7 99.3% leads since October 2026
Not in model cards

Models answer Hebrew questions about everyday life in Israel. The organizers check each answer against an official source and count refusals separately.

Skills IL publishes results for 14 models on 148 questions. The export separates unaided answers, web search, and supplied skills.

Made by
Skills IL
Released
October 2026
Kind
Knowledge in many languages
Level
Everyday
Scored
Checked automatically
Answers
Short answer
Tasks
Public, plus a private test set
Measures
Capability
Language
Hebrew
In and out
Text → Actions, Text
Interaction
Agentic
Environment
Browser
Items from
Written by experts
Refresh
One fixed set
Run by
A host
In views
Agents
Strongest model tested
GPT-6 Astra #4 on the capability index
Last model added
October 2026
Scores refreshed
November 20th, 2026
14
Entries on its own leaderboard, 13 models we track

Grok 4.7 leads with 99.3% among models released by October 2026 Grok 4.7 leads with 99.3% among models released by October 2026

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. Every run is in the next chart. 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 October 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Grok 4.7 With web search 99.3%
  2. 2 Gemini Pro (release unspecified) With skill 99.3%
  3. 3 Gemini 3.8 Flash With skill 98%
  4. 4 Claude Opus 5.5 With skill 97.3%
  5. 5 Kimi K3 With web search 96.6%
  6. 6 Claude Fable 5.1 With skill 96.6%
  7. 7 GPT-6 Astra With web search 96.6%
  8. 8 Claude Sonnet 5.5 With web search 96.6%
  9. 9 GPT-6.1 Sol With web search 96.6%
  10. 10 Qwen 3.8 Max With web search 95.3%
  11. 11 Llama 4 Maverick With skill 94.6%
  12. 12 GLM-5.3 With web search 93.9%
  13. 13 DeepSeek-V4-Pro With skill 93.2%
  14. 14 Mistral Large 4 With web search 92.6%

Source: Official Which AI Model Knows Israel Best? leaderboard

#EntryScore ±SettingReleasedRun onLinkSource
1 Grok 4.7 99.3% 98 to 100 With web search Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
2 Gemini Pro (release unspecified) closed weights 99.3% 98 to 100 With skill Oct 2026 agentskills.co.il Official leaderboard
3 Gemini 3.8 Flash 98% 95.3 to 100 With skill Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
4 Claude Opus 5.5 97.3% 94.6 to 99.3 With skill Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
5 Kimi K3 96.6% 93.2 to 99.3 With web search Jul 2026 Oct 2026 agentskills.co.il Official leaderboard
6 Claude Fable 5.1 96.6% 93.2 to 99.3 With skill Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
7 GPT-6 Astra 96.6% 93.2 to 99.3 With web search Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
8 Claude Sonnet 5.5 96.6% 93.2 to 99.3 With web search Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
9 GPT-6.1 Sol 96.6% 93.2 to 99.3 With web search Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
10 Qwen 3.8 Max 95.3% ±3.4 With web search Aug 2026 Oct 2026 agentskills.co.il Official leaderboard
11 Llama 4 Maverick 94.6% 90.5 to 98 With skill Apr 2025 Oct 2026 agentskills.co.il Official leaderboard
12 GLM-5.3 93.9% 89.9 to 97.3 With web search Aug 2026 Oct 2026 agentskills.co.il Official leaderboard
13 DeepSeek-V4-Pro 93.2% ±4.1 With skill Apr 2026 Oct 2026 agentskills.co.il Official leaderboard
14 Mistral Large 4 92.6% ±4.1 With web search Oct 2026 Oct 2026 agentskills.co.il Official leaderboard
#ModelScore ±SettingReleasedRun onLinkSource
1 Grok 4.7 99.3% 98 to 100 With web search Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
2 Gemini 3.8 Flash 98% 95.3 to 100 With skill Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
3 Claude Opus 5.5 97.3% 94.6 to 99.3 With skill Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
4 Kimi K3 96.6% 93.2 to 99.3 With web search Jul 2026 Oct 2026 agentskills.co.il Official leaderboard
5 Claude Fable 5.1 96.6% 93.2 to 99.3 With skill Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
6 GPT-6 Astra 96.6% 93.2 to 99.3 With web search Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
7 Claude Sonnet 5.5 96.6% 93.2 to 99.3 With web search Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
8 GPT-6.1 Sol 96.6% 93.2 to 99.3 With web search Sep 2026 Oct 2026 agentskills.co.il Official leaderboard
9 Qwen 3.8 Max 95.3% ±3.4 With web search Aug 2026 Oct 2026 agentskills.co.il Official leaderboard
10 Llama 4 Maverick 94.6% 90.5 to 98 With skill Apr 2025 Oct 2026 agentskills.co.il Official leaderboard
11 GLM-5.3 93.9% 89.9 to 97.3 With web search Aug 2026 Oct 2026 agentskills.co.il Official leaderboard
12 DeepSeek-V4-Pro 93.2% ±4.1 With skill Apr 2026 Oct 2026 agentskills.co.il Official leaderboard
13 Mistral Large 4 92.6% ±4.1 With web search Oct 2026 Oct 2026 agentskills.co.il Official leaderboard

Rankings by part

We read 5 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Tax, Employment and National Insurance. GPT-6 Astra leads 4 of them, and GPT-6.1 Sol leads the other one. Arrows show how far a model moved from its overall rank among the models we track.

Tax

13 models we track have a score here.

  1. 1 GPT-6.1 Sol ↑7 96.6%
  2. 2 GPT-6 Astra ↑4 93.1%
  3. 3 Gemini 3.8 Flash ↓1 75.9%
  4. 4 Claude Fable 5.1 ↑1 69%
  5. 5 Claude Opus 5.5 ↓2 69%
  6. 6 Claude Sonnet 5.5 ↑1 65.5%
  7. 7 Kimi K3 ↓3 62.1%
  8. 8 Grok 4.7 ↓7 51.7%
  9. 9 DeepSeek-V4-Pro ↑3 48.3%
  10. 10 Mistral Large 4 ↑3 48.3%

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Grok 4.7
Unaided With skill
99.3%
Gemini 3.8 Flash
Unaided With web se…
98%
Claude Opus 5.5
Unaided With web se…
97.3%
Kimi K3
Unaided With skill
96.6%
Claude Fable 5.1
Unaided With web se…
96.6%
GPT-6 Astra
Unaided With web se…
96.6%
Claude Sonnet 5.5
Unaided With skill
96.6%
GPT-6.1 Sol
Unaided With skill
96.6%
Qwen 3.8 Max
Unaided With skill
95.3%
Llama 4 Maverick
Unaided With web se…
94.6%
GLM-5.3
Unaided With skill
93.9%
DeepSeek-V4-Pro
Unaided With web se…
93.2%

Source: Official Which AI Model Knows Israel Best? leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Grok 4.7 With web search 99.3%
Gemini 3.8 Flash With skill 98%
Gemini 3.8 Flash With web search 97.3%
Claude Opus 5.5 With skill 97.3%
Kimi K3 With web search 96.6%
Claude Fable 5.1 With skill 96.6%
GPT-6 Astra With web search 96.6%
Claude Opus 5.5 With web search 96.6%
Claude Sonnet 5.5 With web search 96.6%
GPT-6.1 Sol With web search 96.6%
Kimi K3 With skill 96%
Grok 4.7 With skill 96%
Qwen 3.8 Max With web search 95.3%
Claude Sonnet 5.5 With skill 95.3%
Llama 4 Maverick With skill 94.6%
Llama 4 Maverick With web search 93.9%
Qwen 3.8 Max With skill 93.9%
GLM-5.3 With web search 93.9%
GPT-6.1 Sol With skill 93.9%
DeepSeek-V4-Pro With skill 93.2%
GPT-6 Astra With skill 93.2%
Claude Fable 5.1 With web search 92.6%
Mistral Large 4 With web search 92.6%
DeepSeek-V4-Pro With web search 91.9%
Mistral Large 4 With skill 91.9%
GLM-5.3 With skill 90.5%
GPT-6 Astra Unaided 87.8%
GPT-6.1 Sol Unaided 86.5%
Gemini 3.8 Flash Unaided 60.8%
Claude Fable 5.1 Unaided 58.8%
Claude Opus 5.5 Unaided 54.7%
Claude Sonnet 5.5 Unaided 50.7%
Kimi K3 Unaided 48%
Grok 4.7 Unaided 44.6%
DeepSeek-V4-Pro Unaided 39.2%
Mistral Large 4 Unaided 35.1%
Qwen 3.8 Max Unaided 33.8%
GLM-5.3 Unaided 31.1%
Llama 4 Maverick Unaided 21.6%

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Grok 4.7
Unaided With skill
99.3%
Gemini 3.8 Flash
Unaided With web se…
98%
Claude Opus 5.5
Unaided With web se…
97.3%
Kimi K3
Unaided With skill
96.6%
Claude Fable 5.1
Unaided With web se…
96.6%
GPT-6 Astra
Unaided With web se…
96.6%
Claude Sonnet 5.5
Unaided With skill
96.6%
GPT-6.1 Sol
Unaided With skill
96.6%
Qwen 3.8 Max
Unaided With skill
95.3%
Llama 4 Maverick
Unaided With web se…
94.6%
GLM-5.3
Unaided With skill
93.9%
DeepSeek-V4-Pro
Unaided With web se…
93.2%

Source: Official Which AI Model Knows Israel Best? leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Grok 4.7 With web search 99.3%
Gemini 3.8 Flash With skill 98%
Gemini 3.8 Flash With web search 97.3%
Claude Opus 5.5 With skill 97.3%
Kimi K3 With web search 96.6%
Claude Fable 5.1 With skill 96.6%
GPT-6 Astra With web search 96.6%
Claude Opus 5.5 With web search 96.6%
Claude Sonnet 5.5 With web search 96.6%
GPT-6.1 Sol With web search 96.6%
Kimi K3 With skill 96%
Grok 4.7 With skill 96%
Qwen 3.8 Max With web search 95.3%
Claude Sonnet 5.5 With skill 95.3%
Llama 4 Maverick With skill 94.6%
Llama 4 Maverick With web search 93.9%
Qwen 3.8 Max With skill 93.9%
GLM-5.3 With web search 93.9%
GPT-6.1 Sol With skill 93.9%
DeepSeek-V4-Pro With skill 93.2%
GPT-6 Astra With skill 93.2%
Claude Fable 5.1 With web search 92.6%
Mistral Large 4 With web search 92.6%
DeepSeek-V4-Pro With web search 91.9%
Mistral Large 4 With skill 91.9%
GLM-5.3 With skill 90.5%
GPT-6 Astra Unaided 87.8%
GPT-6.1 Sol Unaided 86.5%
Gemini 3.8 Flash Unaided 60.8%
Claude Fable 5.1 Unaided 58.8%
Claude Opus 5.5 Unaided 54.7%
Claude Sonnet 5.5 Unaided 50.7%
Kimi K3 Unaided 48%
Grok 4.7 Unaided 44.6%
DeepSeek-V4-Pro Unaided 39.2%
Mistral Large 4 Unaided 35.1%
Qwen 3.8 Max Unaided 33.8%
GLM-5.3 Unaided 31.1%
Llama 4 Maverick Unaided 21.6%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

xAI
Own leaderboard Passed 85% in Oct 2026, by Grok 4.7
Grok 4.7 99.3%

Source: Official Which AI Model Knows Israel Best? leaderboard

Llama 4 Maverick on the official leaderboard passed 85% in April 2025

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.

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Grok 4.7 Llama 4 Maverick 0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Grok 4.7 Llama 4 Maverick

Source: Official Which AI Model Knows Israel Best? 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 Multilingual

All 545 in Multilingual
  • MMMLU In use MMLU professionally translated into 14 languages. Multiple-choice accuracy. Claude Mythos Preview 92.7%
  • WMT24++ In use English-to-55-language translation of WMT24 test data. Scored with automatic metrics such as MetricX or xCOMET. K-EXAONE 90.5%
  • MMLU-ProX In use MMLU-Pro translated into up to 29 languages with expert review. Multiple-choice accuracy per language. Qwen3.7-Max 87%
  • CLIcK In use Models answer 1,995 multiple-choice questions on Korean language and culture. Questions come from official Korean exams and textbooks in eleven categories. Solar Open 2 90.7%
  • Global MMLU Lite 3.0 In use Models answer 400 human-translated MMLU questions per language in 23 languages. Version 3.0 adds Czech, Hungarian, Oriya, Slovak and Tajik and updates Italian. Accuracy. Claude Opus 5 95.7%
  • MMMB In use Models answer multiple-choice questions about images in six languages across 15 categories. The set has about 12,000 questions. LFM2.5-VL-3B 83%

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access Public tasks, plus a private test set.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (14 entries).
  • Pass Models scored 13 models we track, plus 1 other entry on its leaderboard.
  • Pass Newest score The newest is from October 2026, for Mistral Large 4.
  • Pass Frontier models 5 of the top 10 have a score, led by GPT-6 Astra (#4).
  • Note Headroom Saturated. The best is 99.3%, past 85% since April 2025.
Leaderboard import 3 pass
  • Pass Importer We read the board on agentskills.co.il: 39 rows for models we track and 1 other entry 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.

How we check