Benchmark profile

Vals IOI (IOI)

Based on the International Olympiad in Informatics

Data verified July 9, 2026

How BenchLM shows IOI

BenchLM mirrors the public Vals AI IOI leaderboard captured from https://www.vals.ai/benchmarks/ioi and updated by Vals on July 9, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.

IOI is display only on BenchLM. Vals proprietary or Vals-hosted aggregate views are useful context, but BenchLM does not use them as weighted ranking inputs or as a replacement for benchmark-native source records.

58 Vals rows3 task viewspublic datasetTasks: Overall, IOI 2024, IOI 2025Display only

IOI on Vals AI Vals methodology Vals home

IOI score on IOI — July 9, 2026

BenchLM mirrors the published ioi score view for IOI. GPT-5.6 Sol leads the public snapshot at 86.67% , followed by GPT-5.6 Luna (72.92%) and Claude Fable 5 (72.25%). BenchLM does not use these results to rank models overall.

GPT-5.6 Sol

OpenAI

openai/gpt-5.6-sol

86.67%

Overall —

GPT-5.6 Luna

OpenAI

openai/gpt-5.6-luna

72.92%

Overall —

Claude Fable 5

Anthropic

anthropic/claude-fable-5

72.25%

Overall —

58 modelsExternal benchmark mirrorsCurrentDisplay onlyUpdated July 9, 2026

IOI score table (58 models)

Score

GPT-5.6 SolOpenAI

86.67%

GPT-5.6 LunaOpenAI

72.92%

Claude Fable 5Anthropic

72.25%

GPT-5.4OpenAI

67.83%

GPT-5.6 TerraOpenAI

65.25%

GPT-5.2OpenAI

54.83%

Claude Opus 4.7Anthropic

47.08%

Qwen3.7 MaxAlibaba

46.75%

GPT-5.3 CodexOpenAI

43.83%

Gemini 3 Flash PreviewGoogle

39.08%

Gemini 3 Pro PreviewGoogle

38.83%

DeepSeek V4 ProDeepSeek

35.83%

Grok 4.20 0309 ReasoningSpaceXAI

30.17%

Grok 4 0709SpaceXAI

26.17%

Claude Opus 4.5Anthropic

23.58%

GLM 5 ThinkingZhipu AI

22.00%

GPT-5.1OpenAI

21.50%

GPT-5.1 Codex MaxOpenAI

21.42%

Claude Opus 4.5 20251101 ThinkingAnthropic

20.25%

GPT-5OpenAI

20.00%

Claude Sonnet 4.5 20250929 ThinkingAnthropic

18.33%

Kimi K2.5 ThinkingMoonshot AI

17.67%

Gemini 2.5 ProGoogle

17.08%

Qwen3 MaxAlibaba

15.67%

Grok 4.3SpaceXAI

15.33%

GPT-5.4 NanoOpenAI

15.25%

DeepSeek V3p2Fireworks AI

14.42%

Qwen3 MaxAlibaba

13.75%

Claude Opus 4.1Anthropic

12.52%

Grok 4 Fast ReasoningSpaceXAI

11.50%

DeepSeek V3p2 ThinkingFireworks AI

10.67%

GPT-5 CodexOpenAI

9.75%

Qwen3 Max PreviewAlibaba

7.75%

Grok 4.1 Fast Non ReasoningSpaceXAI

7.67%

GLM 4.7Zhipu AI

7.58%

GPT-5 MiniOpenAI

6.75%

MiniMax M2.5MiniMax

6.67%

Claude Sonnet 4Anthropic

6.50%

GPT-5.4 MiniOpenAI

6.42%

Claude Haiku 4.5 20251001 ThinkingAnthropic

6.17%

MiniMax M2.7MiniMax

4.92%

O4 MiniOpenAI

4.83%

Claude Sonnet 4 20250514 ThinkingAnthropic

4.58%

GLM 4.6Zhipu AI

4.33%

Grok Code Fast 1SpaceXAI

4.33%

Mistral Large 2512Mistral AI

4.00%

Grok 4 Fast Non ReasoningSpaceXAI

3.83%

GPT-5.1 CodexOpenAI

3.67%

Grok 4.1 Fast ReasoningSpaceXAI

3.08%

GLM 4.5Zhipu AI

2.92%

Gemini 2.5 FlashGoogle

2.61%

Labs Devstral Small 2512Mistral AI

2.50%

MiniMax M2.1MiniMax

2.33%

DeepSeek V3 0324Fireworks AI

1.67%

Moonshotai Kimi K2 InstructTogether AI

1.25%

Magistral Medium 2509Mistral AI

0.67%

Devstral 2512Mistral AI

0.67%

Qwen3 235b A22bFireworks AI

0.00%

The published IOI snapshot places GPT-5.6 Sol first at 86.67%. The third row is 14.42 points behind. The broader top-10 range is 47.58 points, so the table still separates the published systems.

58 models have been evaluated on IOI. The benchmark falls in the External benchmark mirrors category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. IOI is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About IOI

Year

2026

Tasks

International Olympiad in Informatics-style programming tasks

Format

Accuracy score

Difficulty

Olympiad programming

BenchLM mirrors the public Vals AI IOI leaderboard as display-only external evidence. The captured snapshot preserves overall scores, task-level scores where Vals publishes them, uncertainty, latency, and cost-per-test metadata. It is excluded from BenchLM weighted rankings.

IOI Public benchmark source

BenchLM freshness & provenance

Version

IOI 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

FAQ

What does IOI measure?

Based on the International Olympiad in Informatics

Which model leads the published IOI snapshot?

GPT-5.6 Sol currently leads the published IOI snapshot with 86.67% ioi score. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on IOI?

58 AI models are included in BenchLM's mirrored IOI snapshot, based on the public leaderboard captured on July 9, 2026.

Last updated: July 9, 2026 · mirrored from the public benchmark leaderboard

Choose a model with this week’s evidence

Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.

One email each week. Unsubscribe anytime.