Vals IOI (IOI)
We show this table for reference; we do not rank on it.
Based on the International Olympiad in Informatics
Evaluation results published by Vals AI. Published leaderboard aggregates and evaluation settings only. Dataset contents are not included.
IOI score on IOI — October 29, 2026
We mirror the published ioi score view for IOI. Gemini 4 Argon leads the public snapshot at 100.00%, followed by GPT-6 Astra (100.00%) and GPT-6.1 Sol (96.89%). We do not use these results to rank models overall.
Gemini 4 Argon
high reasoning
GPT-6 Astra
OpenAI
max reasoning
GPT-6.1 Sol
OpenAI
max reasoning
46 modelsCodingCurrentDisplay onlyUpdated October 29, 2026
IOI score table (46 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Gemini 4 ArgonGooglehigh reasoning | 100.00% | Not reported | Closed |
| 2 | GPT-6 AstraOpenAImax reasoning | 100.00% | Not reported | Closed |
| 3 | GPT-6.1 SolOpenAImax reasoning | 96.89% | Not reported | Closed |
| 4 | Claude Opus 5.5Anthropic | 95.06% | Not reported | Closed |
| 5 | GPT-5.6 SolOpenAImax reasoning | 91.17% | Not reported | Closed |
| 6 | Claude Fable 5.1Anthropic | 90.78% | Not reported | Closed |
| 7 | GPT-5.6 TerraOpenAImax reasoning | 87.61% | Not reported | Closed |
| 8 | Claude Opus 5Anthropic | 84.33% | Not reported | Closed |
| 9 | Claude Sonnet 5.5Anthropic | 83.06% | Not reported | Closed |
| 10 | GPT-6 SolOpenAImax reasoning | 82.61% | Not reported | Closed |
| 11 | Kimi K3.1 ProMoonshot AImax reasoning | 81.89% | Not reported | Open |
| 12 | Muse Wonder FullMetamax reasoning | 79.83% | Not reported | Closed |
| 13 | Qwen3.8 MaxAlibaba | 68.89% | Not reported | Open |
| 14 | GLM-5.3Z.AImax reasoning | 68.44% | Not reported | Open |
| 15 | Gemini 3.7 FlashGooglehigh reasoning | 67.83% | Not reported | Closed |
| 16 | Muse WonderMetamax reasoning | 66.94% | Not reported | Closed |
| 17 | GPT-5.6 LunaOpenAImax reasoning | 61.78% | Not reported | Closed |
| 18 | Hy4 previewTencent | 59.33% | Not reported | Open |
| 19 | Grok 4.7xAIxhigh reasoning | 57.72% | Not reported | Closed |
| 20 | Gemini 3.8 FlashGooglehigh reasoning | 56.94% | Not reported | Closed |
| 21 | Muse Spark 1.3 MaxMetamax reasoning | 56.56% | Not reported | Unknown |
| 22 | GPT-6 LunaOpenAImax reasoning | 55.56% | Not reported | Closed |
| 23 | GPT-5.3 CodexOpenAIxhigh reasoning | 53.83% | Not reported | Closed |
| 24 | GLM-5.3-FlashZ.AImax reasoning | 52.50% | Not reported | Open |
| 25 | Gemini 3.1 Pro PreviewGooglehigh reasoning | 51.83% | Not reported | Unknown |
| 26 | DeepSeek V4 Pro 0813DeepSeekmax reasoning | 51.61% | Not reported | Open |
| 27 | Kimi K3Moonshot AImax reasoning | 48.94% | Not reported | Open |
| 28 | MiMo-V2.6-FlashXiaomi | 47.67% | Not reported | Open |
| 29 | Grok 4.6xAIhigh reasoning | 47.61% | Not reported | Closed |
| 30 | Claude Haiku 5.5Anthropic | 47.28% | Not reported | Closed |
| 31 | Ember-1Fireworks | 46.33% | Not reported | Closed |
| 32 | Mistral Large 4Mistralhigh reasoning | 45.28% | Not reported | Open |
| 33 | Claude Sonnet 5Anthropic | 45.00% | Not reported | Closed |
| 34 | Muse Spark 1.3Metaxhigh reasoning | 43.94% | Not reported | Closed |
| 35 | Grok 4.5xAIhigh reasoning | 40.56% | Not reported | Closed |
| 36 | DeepSeek V4.1 Flash 1017DeepSeekmax reasoning | 40.28% | Not reported | Open |
| 37 | MiMo-V2.6-ProXiaomi | 39.33% | Not reported | Open |
| 38 | Qwen3.8-27BAlibabaxhigh reasoning | 39.06% | Not reported | Open |
| 39 | Step 5 PreviewStepFun | 38.06% | Not reported | Open |
| 40 | Gemini 3.6 FlashGooglehigh reasoning | 35.06% | Not reported | Closed |
| 41 | DeepSeek V4 Flash 0731DeepSeekhigh reasoning | 32.72% | Not reported | Open |
| 42 | Beam 1.0Reflection | 22.83% | Not reported | Unknown |
| 43 | Muse Spark 1.2Metaxhigh reasoning | 21.78% | Not reported | Closed |
| 44 | InklingThinking Machines Lab0.99 reasoning | 14.94% | Not reported | Open |
| 45 | Inkling-SmallThinking Machines Lab0.99 reasoning | 9.33% | Not reported | Open |
| 46 | Mercury 2.5Inceptionhigh reasoning | 2.44% | Not reported | Closed |
IOI v1 (Vals, archived)
We retain Vals IOI v1 as an archived, display-only source table. It covers the 2024 and 2025 task sets and stays separate from the live IOI v2 results.
| Source rank | Source model or configuration | IOI v1 score | Source setup | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Claude Opus 5Anthropic | 91.67% | Vals AIOriginal result | Not reported | Closed |
| 2 | GPT-5.6 SolOpenAI | 86.67% | max reasoningOriginal result | Not reported | Closed |
| 3 | Qwen3.8 MaxAlibaba | 73.00% | Vals AIOriginal result | Not reported | Open |
| 4 | GPT-5.6 LunaOpenAI | 72.92% | max reasoningOriginal result | Not reported | Closed |
| 5 | Claude Fable 5Anthropic | 72.25% | Vals AIOriginal result | Not reported | Closed |
| 6 | GPT-5.4OpenAI | 67.83% | xhigh reasoningOriginal result | Not reported | Closed |
| 7 | GPT-5.6 TerraOpenAI | 65.25% | xhigh reasoningOriginal result | Not reported | Closed |
| 8 | GPT-5.2OpenAI | 54.83% | xhigh reasoningOriginal result | Not reported | Closed |
| 9 | Muse Spark 1.2Meta | 49.50% | Vals AIOriginal result | Not reported | Closed |
| 10 | Claude Opus 4.7Anthropic | 47.08% | Vals AIOriginal result | Not reported | Closed |
| 11 | Qwen3.7 MaxAlibaba | 46.75% | Vals AIOriginal result | Not reported | Closed |
| 12 | GPT-5.3 CodexOpenAI | 43.83% | xhigh reasoningOriginal result | Not reported | Closed |
| 13 | Gemini 3 Flash PreviewGoogle | 39.08% | high reasoningOriginal result | Not reported | Unknown |
| 14 | Gemini 3 Pro PreviewGoogle | 38.83% | high reasoningOriginal result | Not reported | Unknown |
| 15 | DeepSeek V4 Pro 0813DeepSeek | 35.83% | max reasoningOriginal result | Not reported | Open |
| 16 | Grok 4.20 0309 ReasoningxAI | 30.17% | Vals AIOriginal result | Not reported | Unknown |
| 17 | Gemini 3.5 Flash-LiteGoogle | 26.17% | high reasoningOriginal result | Not reported | Closed |
| 18 | Grok 4 0709xAI | 26.17% | Vals AIOriginal result | Not reported | Unknown |
| 19 | Claude Opus 4.5Anthropic | 23.58% | Vals AIOriginal result | Not reported | Closed |
| 20 | GLM 5 ThinkingZhipu AI | 22.00% | Vals AIOriginal result | Not reported | Unknown |
| 21 | GPT-5.1OpenAI | 21.50% | high reasoningOriginal result | Not reported | Closed |
| 22 | GPT-5.1-Codex-MaxOpenAI | 21.42% | high reasoningOriginal result | Not reported | Closed |
| 23 | Claude Opus 4.5 ThinkingAnthropic | 20.25% | Vals AIOriginal result | Not reported | Closed |
| 24 | GPT-5OpenAI | 20.00% | high reasoningOriginal result | Not reported | Unknown |
| 25 | Claude Sonnet 4.5 ThinkingAnthropic | 18.33% | Vals AIOriginal result | Not reported | Closed |
| 26 | Kimi K2.5 ThinkingMoonshot AI | 17.67% | Vals AIOriginal result | Not reported | Unknown |
| 27 | Gemini 2.5 ProGoogle | 17.08% | Vals AIOriginal result | Not reported | Closed |
| 28 | Qwen3 MaxAlibaba | 15.67% | Vals AIOriginal result | Not reported | Closed |
| 29 | Grok 4.3xAI | 15.33% | high reasoningOriginal result | Not reported | Closed |
| 30 | GPT-5.4 nanoOpenAI | 15.25% | high reasoningOriginal result | Not reported | Closed |
| 31 | DeepSeek V3p2Fireworks AI | 14.42% | none reasoningOriginal result | Not reported | Unknown |
| 32 | Qwen3 MaxAlibaba | 13.75% | Vals AIOriginal result | Not reported | Closed |
| 33 | Claude Opus 4.1Anthropic | 12.52% | Vals AIOriginal result | Not reported | Unknown |
| 34 | Grok 4 Fast (Reasoning)xAI | 11.50% | Vals AIOriginal result | Not reported | Closed |
| 35 | DeepSeek V3p2 ThinkingFireworks AI | 10.67% | high reasoningOriginal result | Not reported | Unknown |
| 36 | GPT-5 CodexOpenAI | 9.75% | high reasoningOriginal result | Not reported | Unknown |
| 37 | Qwen3 Max PreviewAlibaba | 7.75% | Vals AIOriginal result | Not reported | Unknown |
| 38 | Grok 4.1 Fast Non ReasoningxAI | 7.67% | Vals AIOriginal result | Not reported | Unknown |
| 39 | GLM-4.7Z.AI | 7.58% | Vals AIOriginal result | Not reported | Open |
| 40 | GPT-5 miniOpenAI | 6.75% | high reasoningOriginal result | Not reported | Closed |
| 41 | MiniMax M2.5MiniMax | 6.67% | Vals AIOriginal result | Not reported | Closed |
| 42 | Claude Sonnet 4Anthropic | 6.50% | Vals AIOriginal result | Not reported | Unknown |
| 43 | GPT-5.4 miniOpenAI | 6.42% | xhigh reasoningOriginal result | Not reported | Closed |
| 44 | Claude Haiku 4.5 ThinkingAnthropic | 6.17% | Vals AIOriginal result | Not reported | Closed |
| 45 | MiniMax M2.7MiniMax | 4.92% | Vals AIOriginal result | Not reported | Open |
| 46 | o4-miniOpenAI | 4.83% | high reasoningOriginal result | Not reported | Closed |
| 47 | Claude Sonnet 4 20250514 ThinkingAnthropic | 4.58% | Vals AIOriginal result | Not reported | Unknown |
| 48 | GLM-4.6Z.AI | 4.33% | Vals AIOriginal result | Not reported | Open |
| 49 | Grok Code Fast 1xAI | 4.33% | Vals AIOriginal result | Not reported | Closed |
| 50 | Mistral Large 2512Mistral AI | 4.00% | Vals AIOriginal result | Not reported | Unknown |
| 51 | Grok 4 Fast Non ReasoningxAI | 3.83% | Vals AIOriginal result | Not reported | Unknown |
| 52 | GPT-5.1-CodexOpenAI | 3.67% | high reasoningOriginal result | Not reported | Closed |
| 53 | Grok 4.1 Fast (Reasoning)xAI | 3.08% | Vals AIOriginal result | Not reported | Closed |
| 54 | GLM-4.5Z.AI | 2.92% | Vals AIOriginal result | Not reported | Closed |
| 55 | Gemini 2.5 FlashGoogle | 2.61% | Vals AIOriginal result | Not reported | Closed |
| 56 | Labs Devstral Small 2512Mistral AI | 2.50% | Vals AIOriginal result | Not reported | Unknown |
| 57 | MiniMax M2.1MiniMax | 2.33% | Vals AIOriginal result | Not reported | Unknown |
| 58 | DeepSeek V3 0324DeepSeek | 1.67% | Vals AIOriginal result | Not reported | Open |
| 59 | Moonshotai Kimi K2 InstructTogether AI | 1.25% | Vals AIOriginal result | Not reported | Unknown |
| 60 | Magistral Medium 2509Mistral AI | 0.67% | Vals AIOriginal result | Not reported | Unknown |
| 61 | Devstral 2512Mistral AI | 0.67% | Vals AIOriginal result | Not reported | Unknown |
| 62 | Qwen3 235B-A22BAlibaba | 0.00% | Vals AIOriginal result | Not reported | Open |
How IOI is shown here
BenchLM mirrors the public Vals AI IOI leaderboard captured from https://www.vals.ai/benchmarks/ioi and updated by Vals on October 29, 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.
Snapshot
The published IOI snapshot places Gemini 4 Argon first at 100.00%. The third row is 3.11 points behind. The broader top-10 range is 17.39 points, so the table still separates the published systems.
46 models have been evaluated on IOI. The benchmark falls in the Coding 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.
Freshness and provenance
Version
IOI 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
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.
Questions
What does IOI measure?
Based on the International Olympiad in Informatics
Which model leads the published IOI snapshot?
Gemini 4 Argon currently leads the published IOI snapshot with 100.00% ioi score. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on IOI?
The October 29, 2026 snapshot contains 46 AI models.
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.