Skip to Content Facebook Feature Image

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Business

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents
Business

Business

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

2026-11-18 18:39 Last Updated At:18:55

Top-performing model achieves just 26.7% accuracy, revealing a significant gap in autonomous rule discovery

SINGAPORE, Nov. 18, 2026 /PRNewswire/ -- As AI agents evolve from answering questions to completing complex tasks, a fundamental challenge remains: Can they solve problems when instructions are absent and rules must be discovered entirely from visual clues?

More Images
Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI's Navers Lab today announced Visual Cipher Bench, a new benchmark evaluating long-horizon visual reasoning in AI agents, focusing on rule discovery, logical planning, and spatial operations.

Across eight frontier models and 29 reasoning configurations, the benchmark recorded 3,480 attempts. No configuration achieved accuracy above 30%. The top performer, GPT-6 Astra (max), scored 26.7%. DeepSeek V4.1 Flash, Qwen 3.8 Max, and Kimi K3.1 Pro failed to solve any puzzles.

The findings expose a critical gap between executing known instructions and independently discovering unknown rules. While spatial reasoning proved relatively manageable, accuracy on rule induction was just 0.6%.

Testing AI Beyond Conventional Visual Question Answering

Unlike conventional visual benchmarks, which typically provide explicit questions and expected answers, Visual Cipher Bench tests scenarios where instructions are hidden in images, with no explicit rules or immediate correctness feedback.

The benchmark comprises 40 original visual puzzles represented by 67 black-and-white illustrations. Agents receive one instruction: examine the images and write the final answer to a file. They must independently identify examples, understand the target problem, infer the required output, and solve each task.

The puzzles cover four categories:

  • Spatial Operations: Folding, rotation, projection, and geometric transformations.
  • Logical Planning: Extracting visual constraints and finding valid solutions.
  • Rule Induction: Discovering hidden patterns from limited examples.
  • Knowledge Decoding: Interpreting Morse code, Braille, and word puzzles.

Frontier Models Struggle with Autonomous Rule Discovery

Performance varied significantly across the eight evaluated models:

  • GPT-6 Astra (max): 26.7%, averaging 10.7 correct answers per 40-puzzle run.
  • GPT-6.1 Sol (max) and Claude Opus 5.5 (xhigh): Both scored 21.7%.
  • GPT-6 Luna and Claude Sonnet 5.5: Both scored 0.8%, solving one puzzle each across 120 attempts.
  • DeepSeek V4.1 Flash, Qwen 3.8 Max, and Kimi K3.1 Pro: Zero correct answers.

Across all 3,480 attempts, accuracy by category was:

  • Spatial Operations: 22.8%
  • Knowledge Decoding: 3.3%
  • Logical Planning: 2.4%
  • Rule Induction: 0.6%

Models solved only five of 870 rule-induction attempts. Even GPT-6 Astra (max) achieved 80% on spatial operations, while no configuration exceeded 13.3% in the other three categories. Researchers also found that some correct answers resulted from thematic guesses rather than systematic rule discovery.

Providing Rules Raises Accuracy from 12.2% to 51.1%

To distinguish rule discovery from rule execution, researchers conducted an additional experiment on 10 puzzles using GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna.

When provided with the core rules—but not intermediate reasoning steps or answers—overall accuracy increased from 12.2% to 51.1%. Rule-induction accuracy rose from 0% to 85.2%.

GPT-6 Astra improved from 23.3% to 70.0%, while GPT-6.1 Sol increased from 13.3% to 66.7%. Timeout occurrences also fell from 51 to 29 across 90 attempts.

A control experiment providing additional information about task-state representation, but not the core rules, achieved only 16.7% accuracy. Even with rule hints, logical planning accuracy remained at 22.2%.

Why Do AI Agents Fail?

Analysis of failure trajectories revealed two recurring patterns.

Premature conclusions: Agents often adopted rules that explained only some examples without validating them against all available evidence. Some submitted answers despite acknowledging incomplete understanding.

Inefficient search: Agents frequently relied on extensive tool calls and exhaustive combinations instead of using visual constraints to narrow the search space, eventually timing out.

More effective reasoning involved generating multiple hypotheses, testing them against examples, eliminating inconsistent rules, and applying the remaining valid hypothesis.

From Following Instructions to Discovering Rules

In software development, agents typically receive explicit requirements and can rely on test results to evaluate their solutions. Real-world problems often provide neither.

Agents must observe their environment, identify relevant signals, formulate hypotheses, and validate their understanding without explicit instructions or immediate feedback.

Visual Cipher Bench provides a systematic evaluation of this capability gap. Its findings suggest that independently discovering, validating, and transferring hidden rules remains a significant challenge for frontier AI agents —a capability increasingly important as AI moves toward more autonomous problem-solving.

Research and Resources

** This press release is distributed by PR Newswire through automated distribution system, for which the client assumes full responsibility. **

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

Einsia AI Launches Visual Cipher Bench to Evaluate Long-Horizon Visual Reasoning in AI Agents

The International Tourism Trade Fair will expand its exhibition footprint to eleven halls, reinforcing its position as the global marketplace for the tourism industry.

Specialization remains a core pillar of FITUR, with its 47th edition once again bringing together initiatives focused on sustainability, talent, inclusion and technological transformation.

MADRID, Nov. 18, 2026 /PRNewswire/ -- FITUR 2027, the International Tourism Trade Fair organized by IFEMA MADRID, will take place from 20 to 24 January 2027, with Puerto Rico as Partner Country, in an edition defined by growth, specialization and innovation across the tourism sector. With the addition of two new exhibition halls, the event will further expand its scale and capacity, responding to growing demand from destinations, tourism businesses and industry professionals worldwide.

As the sector continues to evolve, FITUR remains a leading platform for destination promotion, knowledge exchange and business networking. The 2026 edition welcomed 255,000 attendees, including 155,000 industry professionals and 100,000 public visitors. More than 10,000 companies from 161 countries participated, generating an economic impact of €505 million for Madrid.

During the trade days, from 20 to 22 January, experts from around the world will discuss current and future tourism trends, share best practices, present the latest technological tools and establish business agreements. On 23 and 24 January, FITUR will open its doors to the general public, bringing destinations closer to travellers and celebrating tourism in all its forms.

Innovation, AI and Sustainability Take Centre Stage at FITUR 2027

Technological innovation, with AI playing an increasingly central role, alongside sustainability, will once again be among the key themes of FITUR 2027. For the second consecutive year, the Knowledge Hub will serve as the focal point for the discussions, ideas and trends driving transformation in the tourism industry.

Through its program and specialized content areas, FITUR will continue to promote tourism models that generate economic, social and environmental value while encouraging initiatives that support balanced destination development and long-term competitiveness.

FITUR 2027 Continues to Drive Sector Specialization

The exhibition will be organized by geographical regions, including Europe, the Americas, Africa, Asia-Pacific, the Middle East and Spain, alongside a dedicated area for tourism businesses from around the world. It will also once again feature a range of specialized sections supporting key segments of the tourism industry.

These include FITUR Cruises, dedicated to blue tourism and committed to environmental sustainability, destination protection and positive community impact; FITUR LGBT+, a leading platform for one of tourism's most strategic and inclusive segments; FITUR Lingua, bringing together key stakeholders in language tourism; and FITUR Know-How & Export, a platform designed to support the internationalization and knowledge-sharing of tourism companies. In addition, FITUR Screen will showcase the opportunities arising from the relationship between the audiovisual industry and tourism; FITUR Sports will strengthen links between tourism and sport; and FITUR Talent will focus on talent development across the sector.

The event's specialist offering is further enhanced by FITUR Woman, promoting female leadership within tourism; FITUR 4all, dedicated to accessible and inclusive tourism; FITUR Experience, exploring evolving traveller expectations and behaviours; FITUR TechY, focused on technology, innovation and digital transformation; and the FITURNEXT Observatory, which showcases best practices that contribute to a more sustainable tourism industry.

With eleven exhibition halls, Puerto Rico as Partner Country and an increasingly specialized program, FITUR 2027 will offer a global overview of the tourism industry and its key challenges. Innovation, technology, knowledge-sharing and sustainability will define an edition focused on strengthening tourism competitiveness.

For more information, please visit the FITUR website 

MEDIA CONTACT: Elena Valera: evalera@ifema.es. Lucas Farioli: lfarioli@ifema.es 

** This press release is distributed by PR Newswire through automated distribution system, for which the client assumes full responsibility. **

FITUR 2027 to Expand with Two New Halls, Strengthening Its Commitment to Sustainability and Technology

FITUR 2027 to Expand with Two New Halls, Strengthening Its Commitment to Sustainability and Technology

Recommended Articles