Back to Spark
Spark Leaderboard
Live Elo ranking of every model on Spark, computed from anonymous Battle Mode votes. Each vote updates the ratings instantly using the same K-factor Elo system chess uses.
Aug 16, 2026updated2votes cast9models
| Rank | Model | Elo Rating | Win rate | W / L / T | Battles |
|---|---|---|---|---|---|
| #1 | GPT-OSS 20BtoolsOpenAI OSS · via Groq | 1531 | 100% | 2 / 0 / 0 | 2 |
| #2 | Llama 3.3 70B VersatiletoolsMeta · via Groq | 1500 | 0 | ||
| #3 | Llama 3.1 8B InstanttoolsMeta · via Groq | 1500 | 0 | ||
| #4 | Qwen3.6 27BtoolsAlibaba · via Groq | 1500 | 0 | ||
| #5 | Groq CompoundGroq · via Groq | 1500 | 0 | ||
| #6 | Groq Compound MiniGroq · via Groq | 1500 | 0 | ||
| #7 | GLM 5.2toolsZhipu · via OpenRouter | 1500 | 0 | ||
| #8 | Kimi K3toolsMoonshot · via OpenRouter | 1500 | 0 | ||
| #9 | GPT-OSS 120BtoolsOpenAI OSS · via Groq | 1469 | 0% | 0 / 2 / 0 | 2 |
Also on the market
Notable frontier models Spark doesn't run — shown for context only, not benchmarked here.
Claude Opus 5Anthropic
Claude Sonnet 5Anthropic
GPT-5OpenAI
Gemini 3 ProGoogle
DeepSeek V3DeepSeek
Grok 4xAI
Mistral LargeMistral
Phi-4Microsoft