SciCode leaderboard
- Claude Opus 5.5 leads
- 18.5%
- models measured
- 38
- lead over #2
- +4.2 pts
- median
- 9.1%
Leader on SciCode · read October 21
# SciCode, read October 21
leaders = {
"Claude Opus 5.5": 18.5%, "Gemini 3.1 Pro Preview": 14.3%, "Claude Opus 5": 13.8%,
}
On October 21, 2026, 44 of our catalog's models had a SciCode result, and Anthropic Claude Opus 5.5 led with 18.5%, ahead of Google Gemini 3.1 Pro Preview at 14.3%.
SciCode asks a model to write code for real scientific research problems from 16 natural science fields, including mathematics, physics, chemistry and biology. Each of its 80 main problems is split into subproblems. The number is the share of main problems solved without the optional background notes, as Kaggle publishes it.
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 | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | Claude Opus 5.5Anthropic | 18.5% | 88.8 | $20.00 | |
| 2 | Gemini 3.1 Pro PreviewGoogle | 14.3% | 32.3 | $12.00 | |
| 3 | Claude Opus 5Anthropic | 13.8% | 83.0 | $25.00 | |
| 4 | Claude Sonnet 5.5Anthropic | 13.8% | 87.3 | $10.00 | |
| 5 | Gemini 3.7 FlashGoogle | 13.8% | 58.0 | $3.75 | |
| 6 | Gemini 3 Flash PreviewGoogle | 12.7% | 43.7 | $3.00 | |
| 7 | Claude Opus 4.8Anthropic | 12.3% | 65.0 | $25.00 | |
| 8 | Gemini 3.5 FlashGoogle | 12.3% | 39.8 | $9.00 | |
| 9 | Gemini 3.8 FlashGoogle | 12.3% | 61.6 | $3.75 | |
| 10 | Gemini 4 ArgonGoogle | 12.3% | 81.0 | $10.00 | |
| 11 | GPT-6 AstraOpenAI | 12.3% | 85.1 | $50.00 | |
| 12 | Gemini 3.6 FlashGoogle | 10.8% | 62.5 | $3.75 | |
| 13 | GPT-5.6 SolOpenAI | 10.8% | 73.5 | $20.00 | |
| 14 | GPT-5.6 TerraOpenAI | 10.8% | 63.9 | $12.00 | |
| 15 | o4 MiniOpenAI | 10.8% | Not benchmarked | $4.40 | |
| 16 | Claude Sonnet 5Anthropic | 10.7% | 57.7 | $10.00 | |
| 17 | Claude Haiku 5.5Anthropic | 9.2% | 70.7 | $0.50 | |
| 18 | GPT-6.1 SolOpenAI | 9.2% | 82.4 | $10.00 | |
| 19 | o3OpenAI | 9.2% | Not benchmarked | $8.00 | |
| 20 | Gemma 4 31BGoogle | 8.9% | 39.0 | $0.40 | |
| 21 | Gemini 2.5 Pro Preview 06-05Google | 7.7% | Not benchmarked | $10.00 | |
| 22 | gpt-oss-120bOpenAI | 7.7% | Not benchmarked | $0.50 | |
| 23 | o3 MiniOpenAI | 7.7% | Not benchmarked | $4.40 | |
| 24 | Grok 4.5xAI | 6.5% | 58.6 | $6.00 | |
| 25 | Claude Opus 4RetiredAnthropic | 6.2% | Not benchmarked | $75.00 | |
| 26 | Gemini 3.5 Flash LiteGoogle | 6.2% | 49.0 | $2.50 | |
| 27 | GPT-5.6 LunaOpenAI | 6.2% | 58.0 | $1.20 | |
| 28 | Claude 3.7 SonnetRetiredAnthropic | 4.6% | Not benchmarked | $15.00 | |
| 29 | Claude Opus 4.1Anthropic | 4.6% | Not benchmarked | $75.00 | |
| 30 | R1RetiredDeepSeek | 4.6% | Not benchmarked | $2.50 | |
| 31 | Gemini 3.1 Flash Lite PreviewGoogle | 4.6% | 22.7 | $1.50 | |
| 32 | GPT-4.1OpenAI | 4.6% | Not benchmarked | $8.00 | |
| 33 | o1OpenAI | 4.6% | Not benchmarked | $60.00 | |
| 34 | o1-miniRetiredOpenAI | 3.1% | Not benchmarked | $4.40 | |
| 35 | Claude 3.5 HaikuRetiredAnthropic | 1.5% | Not benchmarked | $4.00 | |
| 36 | GPT-4oOpenAI | 1.5% | Not benchmarked | $10.00 | |
| 37 | GPT-4o-miniOpenAI | 1.5% | Not benchmarked | $0.60 | |
| 38 | Grok 3RetiredxAI | 1.5% | Not benchmarked | $15.00 | |
| 39 | Gemma 3 12BGoogle | 0.0% | Not benchmarked | $0.1625 | |
| 40 | Gemma 3 27BGoogle | 0.0% | Not benchmarked | $0.30 | |
| 41 | Gemma 3 4BGoogle | 0.0% | Not benchmarked | $0.075 | |
| 42 | Mixtral 8x22B InstructMistral AI | 0.0% | Not benchmarked | $6.00 | |
| 43 | GPT-3.5 TurboOpenAI | 0.0% | Not benchmarked | $1.50 | |
| 44 | GPT-5OpenAI | 0.0% | 38.8 | $10.00 |
Only models in our catalog are listed, and each appears once with its best published run. 59 rows were read from the source on October 21, 2026, and rows that match no catalog model are left out.
| Model | Result | Main problems, with background | Subproblems | Subproblems, with background |
|---|---|---|---|---|
| Anthropic Claude Opus 5.5 | 18.5% | 21.5% | 52.2% | 58.1% |
| Google Gemini 3.1 Pro Preview | 14.3% | 30.0% | 44.9% | 50.0% |
| Anthropic Claude Opus 5 | 13.8% | 20.0% | 42.6% | 49.5% |
| Anthropic Claude Sonnet 5.5 | 13.8% | 18.5% | 43.3% | 50.2% |
| Google Gemini 3.7 Flash | 13.8% | 21.5% | 43.0% | 54.3% |
| Google Gemini 3 Flash Preview | 12.7% | 16.9% | 30.9% | 47.1% |
| Anthropic Claude Opus 4.8 | 12.3% | 16.9% | 38.1% | 45.0% |
| Google Gemini 3.5 Flash | 12.3% | 18.5% | 39.5% | 49.1% |
| Google Gemini 3.8 Flash | 12.3% | 20.0% | 40.2% | 54.6% |
| Google Gemini 4 Argon | 12.3% | 18.5% | 45.0% | 55.3% |
| OpenAI GPT-6 Astra | 12.3% | 18.5% | 43.3% | 56.4% |
| Google Gemini 3.6 Flash | 10.8% | 16.9% | 37.8% | 49.5% |
| OpenAI GPT-5.6 Sol | 10.8% | 23.1% | 42.6% | 54.6% |
| OpenAI GPT-5.6 Terra | 10.8% | 18.5% | 39.2% | 48.5% |
| OpenAI o4 Mini | 10.8% | 9.2% | 36.1% | 39.9% |
| Anthropic Claude Sonnet 5 | 10.7% | 21.7% | 40.1% | 49.3% |
| Anthropic Claude Haiku 5.5 | 9.2% | 10.8% | 34.4% | 41.6% |
| OpenAI GPT-6.1 Sol | 9.2% | 18.5% | 37.8% | 53.3% |
| OpenAI o3 | 9.2% | 9.2% | 36.4% | 44.7% |
| Google Gemma 4 31B | 8.9% | 9.8% | 33.2% | 42.9% |
| Google Gemini 2.5 Pro Preview 06-05 | 7.7% | 15.4% | 30.9% | 41.2% |
| OpenAI gpt-oss-120b | 7.7% | 6.2% | 27.5% | 34.0% |
| OpenAI o3 Mini | 7.7% | 12.3% | 34.7% | 43.3% |
| SpaceXAI Grok 4.5 | 6.5% | 21.5% | 41.9% | 50.2% |
| Anthropic Claude Opus 4Retired | 6.2% | 12.3% | 30.6% | 41.9% |
| Google Gemini 3.5 Flash Lite | 6.2% | 12.3% | 29.6% | 38.1% |
| OpenAI GPT-5.6 Luna | 6.2% | 12.3% | 34.4% | 41.9% |
| Anthropic Claude 3.7 SonnetRetired | 4.6% | 7.7% | 30.6% | 38.8% |
| Anthropic Claude Opus 4.1 | 4.6% | 9.2% | 27.1% | 38.1% |
| DeepSeek R1Retired | 4.6% | 10.8% | 29.6% | 37.8% |
| Google Gemini 3.1 Flash Lite Preview | 4.6% | 6.2% | 27.5% | 34.0% |
| OpenAI GPT-4.1 | 4.6% | 9.2% | 30.9% | 38.1% |
| OpenAI o1 | 4.6% | 10.8% | 29.2% | 38.5% |
| OpenAI o1-miniRetired | 3.1% | 3.1% | 25.1% | 36.1% |
| Anthropic Claude 3.5 HaikuRetired | 1.5% | 0.0% | 22.0% | 30.6% |
| OpenAI GPT-4o | 1.5% | 7.7% | 24.1% | 33.7% |
| OpenAI GPT-4o-mini | 1.5% | 0.0% | 16.5% | 24.1% |
| xAI Grok 3Retired | 1.5% | 6.2% | 29.2% | 38.1% |
| Google Gemma 3 12B | 0.0% | 0.0% | 12.0% | 18.6% |
| Google Gemma 3 27B | 0.0% | 0.0% | 14.8% | 23.4% |
| Google Gemma 3 4B | 0.0% | 0.0% | 4.8% | 9.3% |
| Mistral Mixtral 8x22B Instruct | 0.0% | 3.1% | 15.5% | 22.3% |
| OpenAI GPT-3.5 Turbo | 0.0% | 0.0% | 13.1% | 18.6% |
| OpenAI GPT-5 | 0.0% | 0.0% | 33.3% | 29.6% |
The parts as Kaggle Benchmarks publishes them, for the same run as the result.