AI went from about 4% to 95% on puzzles people can solve, and under 1% to 81% on puzzle games
Puzzles ordinary people can solve stumped AI in early 2025. By September 2026 the best model solved 95%, checked by the ARC Prize team.
Updated 10 Nov 2026: Claude Fable 5.2 raised the best ARC-AGI-3 standard-harness score from 62.7% to 81.2%.
ARC-AGI-2 puzzles: with and without a harness
Best ARC Prize-verified score on the semi-private set, by model release date.
What this measures: Visual logic puzzles that every task was solved by at least two people, but that AI found almost impossible in early 2025. A "harness" is extra code around the model that lets it try, check and refine answers.
- Model alone (ARC's standard setup)
- With a third-party refinement harness
- Reference level
Model alone (ARC's standard setup), Saturated: The best model scores 95% (September 2026), and ARC Prize says ARC-AGI-1 and -2 are being overfit. Its focus has moved to the interactive ARC-AGI-3. Now watch: ARC-AGI-3.
With a third-party refinement harness, No new results · last data Feb 2026: No new verified harness result since February 2026: plain models overtook the harness systems (GPT-6 Astra scored 95% at $1.12 per task, versus 72.9% at $38.99 for the best harness), so there has been little reason to submit new ones.
Harness systems led in 2025, then plain models overtook them in 2026. The compute-capped ARC Prize 2025 Kaggle competition was won by NVARC with 24.0% at about $0.20 per task. ARC Prize has said ARC-AGI-1 and -2 are being overfit, intentionally or not.
Show data table and sources
| Series | Date | Label | Value |
|---|---|---|---|
| Model alone (ARC's standard setup) | 20 Dec 2024 | o3-preview (low) | 4% |
| Model alone (ARC's standard setup) | 16 Apr 2025 | o3 (high) | 6.5% |
| Model alone (ARC's standard setup) | 14 May 2025 | Claude Opus 4 | 8.6% |
| Model alone (ARC's standard setup) | 9 Jul 2025 | Grok 4 (thinking) | 16% |
| Model alone (ARC's standard setup) | 6 Oct 2025 | GPT-5 Pro | 18% |
| Model alone (ARC's standard setup) | 18 Nov 2025 | Gemini 3 Pro | 31% |
| Model alone (ARC's standard setup) | 24 Nov 2025 | Claude Opus 4.5 | 38% |
| Model alone (ARC's standard setup) | 11 Dec 2025 | GPT-5.2 Pro (high) | 54% |
| Model alone (ARC's standard setup) | 5 Feb 2026 | Claude Opus 4.6 (high) | 69% |
| Model alone (ARC's standard setup) | 12 Feb 2026 | Gemini 3 Deep Think | 85% |
| Model alone (ARC's standard setup) | 22 Apr 2026 | GPT-5.5 (xhigh) | 85% |
| Model alone (ARC's standard setup) | 9 Jun 2026 | Claude Fable 5 (max) | 89% |
| Model alone (ARC's standard setup) | 9 Jul 2026 | GPT-5.6 Sol (max) | 93% |
| Model alone (ARC's standard setup) | 2 Sept 2026 | GPT-6 Astra (max) | 95% |
| With a third-party refinement harness | 7 Aug 2025 | Grok 4 + refinement (j_berman) | 29% |
| With a third-party refinement harness | 4 Dec 2025 | Gemini 3 Pro + Poetiq | 54% |
| With a third-party refinement harness | 3 Feb 2026 | GPT-5.2 + refinement (johan_land) | 73% |
| Reference | Human panel (100%) | 100% | |
| Reference | Average test-taker (60%) | 60% |
- ARC-AGI Leaderboard (ARC Prize)
- ARC Prize leaderboard data (evaluations.json) (ARC Prize)
- Announcing ARC-AGI-2 and ARC Prize 2025 (ARC Prize)
- ARC Prize 2025 Results & Analysis (ARC Prize)
ARC-AGI-3: interactive puzzle games
Best ARC Prize score on the semi-private set. Standard harness vs a lab's own harness.
What this measures: Hundreds of small turn-based games with no instructions: the AI has to explore, work out the rules and win in as few moves as possible. Every game was beaten by human testers; 100% means matching a typical first-time human player's efficiency on every game.
- ARC's standard harness
- Lab's own harness (provider adapter)
- Reference level
The harness matters on interactive tasks: every model tested both ways scored far higher when it could keep its own hidden reasoning between moves (e.g. GPT-6 Astra 62.7% → 99.9%, Claude Fable 5.2 81.2% → 97.7%). Dates are model release dates. ARC-AGI-3 costs are per run, not per task. ARC Prize 2026 competition results are due on 4 December 2026.
Show data table and sources
| Series | Date | Model | Value |
|---|---|---|---|
| ARC's standard harness | 25 Mar 2026 | Best frontier AI at launch | 0.5% |
| ARC's standard harness | 1 Jun 2026 | Claude Opus 4.8 (high) | 1.5% |
| ARC's standard harness | 9 Jul 2026 | GPT-5.6 Sol (max) | 7.8% |
| ARC's standard harness | 24 Jul 2026 | Claude Opus 5 (high) | 30% |
| ARC's standard harness | 2 Sept 2026 | GPT-6 Astra (max) | 63% |
| ARC's standard harness | 2 Sept 2026 | Gemini 3.8 Flash | 10% |
| ARC's standard harness | 28 Sept 2026 | Claude Sonnet 5.5 | 24% |
| ARC's standard harness | 29 Sept 2026 | GPT-6.1 Sol | 53% |
| ARC's standard harness | 30 Sept 2026 | Gemini 4 Argon | 50% |
| ARC's standard harness | 9 Nov 2026 | Claude Fable 5.2 | 81% |
| Lab's own harness (provider adapter) | 2 Sept 2026 | Gemini 3.8 Flash + adapter | 35% |
| Lab's own harness (provider adapter) | 3 Sept 2026 | GPT-6 Astra + OpenAI adapter | 100% |
| Lab's own harness (provider adapter) | 28 Sept 2026 | Claude Sonnet 5.5 + adapter | 72% |
| Lab's own harness (provider adapter) | 29 Sept 2026 | GPT-6.1 Sol + adapter | 96% |
| Lab's own harness (provider adapter) | 30 Sept 2026 | Gemini 4 Argon + adapter | 98% |
| Lab's own harness (provider adapter) | 9 Nov 2026 | Claude Fable 5.2 + adapter | 98% |
| Reference | Typical first-time human player (100%) | 100% |
- ARC-AGI Leaderboard (ARC Prize)
- ARC Prize leaderboard data (evaluations.json) (ARC Prize)
- ARC-AGI-3 launch (ARC Prize)
- GPT-6 Astra results (ARC Prize)
- OpenAI's GPT-6 Astra on ARC-AGI-3 (ARC Prize Foundation)
ARC Prize, the group behind the ARC-AGI tests, builds puzzles that people find manageable and AI has found hard. It checks the top scores itself, on a set of puzzles that is not public.
ARC-AGI-2 is a set of visual pattern puzzles. Every puzzle was solved by at least two people in two tries or fewer, and the average test-taker scored 60%. At its launch in March 2025, ARC Prize estimated OpenAI’s o3-preview at about 4%. Verified scores then rose to 54.2% (GPT-5.2 Pro, December 2025) and 95.0% (GPT-6 Astra, September 2026, at $1.12 per puzzle). In ARC Prize’s own 2025 Kaggle competition, where computing power is capped and solutions must be open source, the winner scored 24.03%.
ARC-AGI-3, launched in March 2026, is hundreds of turn-based games with no instructions. The AI has to work out the rules by playing. It is scored on how efficiently it acts: 100% means matching a typical first-time human player on every game. Human testers beat every game. The best AI scored 0.51% at launch.
On these games the harness matters (the software that lets a model act in the game and keep notes between turns). With ARC Prize’s standard harness, the best score rose to 30.2% (Claude Opus 5, July 2026), 62.7% (GPT-6 Astra, September) and 81.2% (Claude Fable 5.2, November). With OpenAI’s own harness, which keeps the model’s reasoning from one turn to the next, GPT-6 Astra scored 98.6–99.9%. That is a separate result, not comparable with the standard ones.
What it means
Puzzles designed to show what AI could not do went from nearly unsolved to nearly solved in under two years. Claims that AI fails at simple puzzles go out of date quickly. ARC Prize does not treat this as proof of human-level general intelligence, and it suspects some scores are helped by models being tuned to its puzzles.
Caveats
- In December 2025 ARC Prize said it believes ARC-AGI-1 and -2 “are being overfit, either incidentally or intentionally”, meaning models are tuned to the puzzles. It cited a Gemini 3 run that used ARC’s colour codes without being told about ARC.
- Lab runs use far more computing power than the Kaggle competition allows; ARC Prize treats cost per puzzle as part of the result.
- Some top ARC-AGI-2 entries are systems built by third parties on top of a model, not a bare model.
- Solving a version is not a claim of AGI (AI that matches people across the board). ARC Prize frames each version as measuring a remaining gap between AI and people.