Kaggle Gemma 4 Good Hackathon — Verified Facts

Deadline

• Final submission deadline: May 18, 2026, 23:59 UTC.

  "$200K Prize Pool Final Submission: May 18, 2026"

  Source 1: Kaggle official X (Twitter) announcement — x.com/kaggle/status/2039740198259462370

  "The deadline is May 18, 2026, 23:59 UTC."

  Source 2: Algo-Mania write-up — algo-mania.com/.../gemma-4-hackathon-200000-to-create-ai-for-social-impact
  Source 3: Kaggle official LinkedIn post — linkedin.com/posts/kaggle_...activity-7445505784228073472 ("Submission Deadline: May 18, 2026")
• Competition opened April 2, 2026.
  Source 1: johnsonhk88 participant repo README — github.com/johnsonhk88/Kaggle-The-Gemma-4-Good-Hackathon ("Start Date: April 2, 2026")
  Source 2: EdTech Innovation Hub coverage of launch — edtechinnovationhub.com/news/kaggle-and-google-deepmind-open-gemma-4-hackathon

Prize Pool

• Total prize pool: $200,000 USD.

  "Total Prize Pool: $200,000"

  Source 1: Kaggle official LinkedIn post — linkedin.com/posts/kaggle_...activity-7445505784228073472

  "$200K Prize Pool"

  Source 2: Kaggle official X (Twitter) announcement — x.com/kaggle/status/2039740198259462370

  "There's $200,000 total prizes to be won."

  Source 3: Unsloth AI X (Twitter) — x.com/UnslothAI/status/2042599142560796991
• Awards split across three category tiers: general, impact-focused, and technical.

  "$200,000 total, with awards across general, impact-focused, and technical categories"

  Source 1: Algo-Mania — algo-mania.com
  Source 2: EdTech Innovation Hub — edtechinnovationhub.com
• Special $10,000 Unsloth prize for best fine-tuned Gemma 4 model built with Unsloth.

  "Google DeepMind is hosting a Gemma 4 hackathon with a $10,000 Unsloth prize! Show off your best fine-tuned Gemma 4 model built with Unsloth."

  Source 1: Unsloth AI official X (Twitter) — x.com/UnslothAI/status/2042599142560796991
  Source 2: Algo-Mania — algo-mania.com ("special mentions for Ollama and Unsloth technologies")

Tracks

• Five impact-focused tracks. Names and descriptions below match across the Kaggle Team author's Medium post and independent press coverage.

  "1. Health & Sciences — Tools that simplify medical/scientific information for frontline workers.
  2. Global Resilience — Systems addressing crises, infrastructure gaps, and disaster response.
  3. Future of Education — Solutions for underserved learning environments with language or connectivity barriers.
  4. Digital Equity & Inclusivity — Tools reducing barriers for marginalized or non-technical users.
  5. Safety & Trust — Systems improving hallucination reduction and verification reliability."

  Source 1: "The Gemma 4 Good Hackathon" by Kaggle Team author Sudha Rani Maddala on Medium — sudhamsr.medium.com/the-gemma-4-good-hackathon-aef927f17ef1

  "The five focus areas are: Future of Education, Health and Sciences, Digital Equity, Global Resilience, Safety."

  Source 2: The Inner Detail — theinnerdetail.com/google-announces-hackathon-on-ai-skills-using-gemma-4-with-200k-prize/
  Source 3: EdTech Innovation Hub — edtechinnovationhub.com (same five tracks)
• Within Global Resilience there is a named focus on Climate & Green Energy.

  "special tracks for Global Resilience – Climate & Green Energy"

  Source 1: Algo-Mania — algo-mania.com

  "solve real-world challenges in health, education and climate"

  Source 2: Kaggle official X — x.com/kaggle/status/2039740198259462370

Eligibility

• A Kaggle account with completed identity verification is required to enter.

  "a Kaggle account, identity verification completed"

  Source 1: Kaggle Team author article on Medium — sudhamsr.medium.com/the-gemma-4-good-hackathon-aef927f17ef1
  Source 2: EdTech Innovation Hub — edtechinnovationhub.com
• Open globally; framed as a "global hackathon."

  "a global challenge designed to foster real-world AI applications"

  Source 1: EdTech Innovation Hub — edtechinnovationhub.com
  Source 2: Algo-Mania — algo-mania.com ("global competition")

Submission

• Submissions must use at least one Gemma 4 model.

  "Solutions must use at least one Gemma 4 model."

  Source 1: Algo-Mania — algo-mania.com
  Source 2: EdTech Innovation Hub — edtechinnovationhub.com
• Available Gemma 4 model sizes: 31B Dense, 26B MoE, and two Edge variants E4B and E2B. Released under Apache 2.0.

  "Participants can download Gemma 4 in four sizes: two high-performance versions (31B Dense & 26B MoE) and two optimized Edge versions (E4B & E2B) ... under an Apache 2.0 License."

  Source 1: Kaggle official LinkedIn post — linkedin.com/posts/kaggle_...activity-7445505784228073472

  "Projects are released under an Apache 2.0 license"

  Source 2: The Inner Detail — theinnerdetail.com

Deliverables

• Required submission package: working demo, public code repository, technical write-up, public video, and a cover image / media gallery.

  "a Kaggle account, identity verification completed, a public project write-up, a public code repository, a public demo or demo files, a public video, a cover image / media gallery assets"

  Source 1: Kaggle Team author article on Medium — sudhamsr.medium.com/the-gemma-4-good-hackathon-aef927f17ef1

  "Submissions must include a working demo, a public code repository, and a technical write-up outlining how Gemma 4 has been applied, alongside a short video demonstrating real-world use."

  Source 2: EdTech Innovation Hub — edtechinnovationhub.com
  Source 3: Algo-Mania — algo-mania.com (same four artifacts)
• The technical write-up must specifically describe how Gemma 4 features were used.

  "a technical write-up detailing how the specific features of Gemma 4 were implemented"

  Source 1: The Inner Detail — theinnerdetail.com
  Source 2: Algo-Mania — algo-mania.com
• Submission profile resembles a product launch, not a notebook drop.

  "this resembles a mini product launch rather than a casual notebook submission"

  Source: Kaggle Team author article on Medium — sudhamsr.medium.com/the-gemma-4-good-hackathon-aef927f17ef1

Judging

• Three judging axes: Impact & Vision, Video Pitch & Storytelling, and Technical Depth & Execution.

  "Judges evaluate: Impact & Vision (authentic problem identification); Video Pitch & Storytelling (clear communication); Technical Depth & Execution (working proof-of-concept)"

  Source 1: Kaggle Team author article on Medium — sudhamsr.medium.com/the-gemma-4-good-hackathon-aef927f17ef1

  "Impact & Vision, Video Pitch & Storytelling, Technical Depth & Execution"

  Source 2: EdTech Innovation Hub — edtechinnovationhub.com
• The judging committee explicitly does not reward complexity for its own sake; clarity of problem definition and execution win.

  "winners won't necessarily have the most complicated architecture"

  Source 1: Kaggle Team author article on Medium — sudhamsr.medium.com/the-gemma-4-good-hackathon-aef927f17ef1

  "Judging will be heavily weighted toward technical execution and the ability of the project to solve a concrete problem."

  Source 2: The Inner Detail — theinnerdetail.com

Technical Hooks Encouraged

• Two Gemma 4 capabilities the organizers explicitly call out: multimodal understanding and native function calling.

  "Use multimodal power and native function calling to solve real-world challenges in health, education and climate."

  Source 1: Kaggle official X — x.com/kaggle/status/2039740198259462370

  "local frontier intelligence, native function calling, and multimodal understanding"

  Source 2: Kaggle official LinkedIn post — linkedin.com/posts/kaggle_...activity-7445505784228073472