IOI (Vals AI) leaderboard
- Gemini 4 Argon leads
- 100.0%
- models measured
- 42
- at the top
- Tied
- median
- 52.2%
Leader on IOI (Vals AI) · read October 22
# IOI (Vals AI), read October 22
leaders = {
"Gemini 4 Argon": 100.0%, "GPT-6 Astra": 100.0%, "GPT-6.1 Sol": 96.9%,
}
On October 22, 2026, 42 of our catalog's models had a IOI (Vals AI) result, and Google Gemini 4 Argon led with 100.0%.
Problems from the International Olympiad in Informatics, run by Vals AI. Share of points.
Role in the index
A coding board we show beside the Vibe Coding Index, not inside it: the index keeps the six boards it was defined on, so a score does not move every time a board is added. Read the methodology
| # | Model | Result | Cost per test | Duration, as printed | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|---|---|
| 1 | Gemini 4 ArgonGoogle | 100.0% | $5.31 | 24m 19s | 81.0 | $10.00 | |
| 2 | GPT-6 AstraOpenAI | 100.0% | $6.50 | 31m 40s | 85.1 | $50.00 | |
| 3 | GPT-6.1 SolOpenAI | 96.9% | $1.26 | 31m 31s | 82.4 | $10.00 | |
| 4 | Claude Opus 5.5Anthropic | 95.1% | $5.26 | 17m 15s | 88.8 | $20.00 | |
| 5 | GPT-5.6 SolOpenAI | 91.2% | $7.98 | 1h 02m | 73.5 | $20.00 | |
| 6 | Claude Fable 5.1Anthropic | 90.8% | $11.20 | 30m 06s | 84.2 | $50.00 | |
| 7 | GPT-5.6 TerraOpenAI | 87.6% | $8.66 | 1h 23m | 63.9 | $12.00 | |
| 8 | Claude Opus 5Anthropic | 84.3% | $16.48 | 56m 42s | 83.0 | $25.00 | |
| 9 | Claude Sonnet 5.5Anthropic | 83.1% | $7.52 | 45m 51s | 87.3 | $10.00 | |
| 10 | GPT-6 SolOpenAI | 82.6% | $2.70 | 30m 38s | 76.0 | $10.00 | |
| 11 | Qwen3.8 Max (0902)Qwen | 68.9% | $9.47 | 1h 47m | 62.4 | $6.00 | |
| 12 | GLM 5.3Z.ai | 68.4% | $7.67 | 1h 46m | 70.6 | $4.40 | |
| 13 | Gemini 3.7 FlashGoogle | 67.8% | $3.40 | 21m 09s | 58.0 | $3.75 | |
| 14 | Muse WonderMeta | 66.9% | $4.54 | 25m 02s | 71.9 | $20.00 | |
| 15 | GPT-5.6 LunaOpenAI | 61.8% | $0.58 | 1h 10m | 58.0 | $1.20 | |
| 16 | Hy4 previewTencent | 59.3% | $1.35 | 1h 01m | 44.3 | $2.501 | |
| 17 | Grok 4.7xAI | 57.7% | $12.71 | 46m 48s | 66.7 | $6.00 | |
| 18 | Gemini 3.8 FlashGoogle | 56.9% | $3.98 | 14m 20s | 61.6 | $3.75 | |
| 19 | GPT-6 LunaOpenAI | 55.6% | $0.20 | 35m 47s | 57.3 | $0.50 | |
| 20 | GPT-5.3-CodexOpenAI | 53.8% | $2.68 | 29m 52s | 60.0 | $14.00 | |
| 21 | GLM 5.3 FlashZ.ai | 52.5% | $0.43 | 1h 32m | 57.7 | $0.50 | |
| 22 | Gemini 3.1 Pro PreviewGoogle | 51.8% | $1.25 | 12m 03s | 32.3 | $12.00 | |
| 23 | DeepSeek V4 Pro 0813DeepSeek | 51.6% | $2.15 | 1h 07m | 56.1 | $1.98 | |
| 24 | Kimi K3Moonshot AI | 48.9% | $14.17 | 4h 18m | 62.3 | $15.00 | |
| 25 | MiMo-V2.6-FlashXiaomi | 47.7% | $0.22 | 1h 49m | 54.7 | $0.28 | |
| 26 | Grok 4.6xAI | 47.6% | $8.18 | 3h 57m | 65.7 | $6.00 | |
| 27 | Claude Haiku 5.5Anthropic | 47.3% | $0.83 | 12m 34s | 70.7 | $0.50 | |
| 28 | Ember-1Fireworks | 46.3% | $24.46 | 1h 47m | 52.6 | $15.00 | |
| 29 | Mistral Large 4Mistral AI | 45.3% | $5.39 | 1h 24m | 56.4 | $4.18 | |
| 30 | Claude Sonnet 5Anthropic | 45.0% | $12.89 | 1h 00m | 57.7 | $10.00 | |
| 31 | Muse Spark 1.3Meta | 43.9% | $2.36 | 29m 51s | 57.4 | $4.25 | |
| 32 | Grok 4.5xAI | 40.6% | $5.64 | 1h 27m | 58.6 | $6.00 | |
| 33 | DeepSeek V4.1 FlashDeepSeek | 40.3% | $0.24 | 18m 30s | 62.5 | $0.84 | |
| 34 | MiMo-V2.6-ProXiaomi | 39.3% | $0.84 | 2h 26m | 65.4 | $0.87 | |
| 35 | Qwen3.8 27BQwen | 39.1% | $3.55 | 1h 39m | 74.2 | $2.55 | |
| 36 | Step 5 PreviewStepFun | 38.1% | $2.03 | 1h 20m | 61.1 | $2.70 | |
| 37 | Gemini 3.6 FlashGoogle | 35.1% | $8.67 | 48m 23s | 62.5 | $3.75 | |
| 38 | DeepSeek V4 Flash 0731DeepSeek | 32.7% | $0.37 | 27m 43s | 51.8 | $0.396 | |
| 39 | Muse Spark 1.2Meta | 21.8% | $2.64 | 25m 35s | 55.7 | $4.25 | |
| 40 | T InklingThinking Machines | 14.9% | $1.71 | 30m 12s | 31.5 | $4.05 | |
| 41 | T Inkling SmallThinking Machines | 9.3% | $0.34 | 11m 18s | 27.3 | $1.20 | |
| 42 | Mercury 2.5Inception | 2.4% | $0.16 | 4m 03s | 1.6 | $0.15 |
Only models in our catalog are listed, and each appears once with its best published run. 137 rows were read from the source on October 22, 2026, and rows that match no catalog model are left out. Cost per test is what Vals AI reports one test cost on the run behind the result, in US dollars. Every other column is a figure the board publishes for the same run.