| Metric | Value |
|---|---|
| Firebase wins | 744 |
| Ably wins | 929 |
| Abstains (no tool) | 42 |
| Other tool chosen | 2021 |
| Decisive cases | 1673 |
| Firebase win rate (unweighted) | 44.5% |
| 95% CI | 42.1% - 46.9% |
| Firebase win rate (weighted) | 44.5% |
Verified critics can leave comments here.
Verified critics can leave comments here.
| Model | Tier | Firebase | Ably | None | Other | A rate |
|---|---|---|---|---|---|---|
| Devstral 2 2512 | Mid | 195 | 0 | 0 | 3 | 100% |
| Llama 4 Scout | Small | 60 | 62 | 0 | 73 | 49% |
| GPT 5.3 Codex | Frontier | 0 | 120 | 0 | 78 | 0% |
| Mistral Small 4 | Mid | 79 | 39 | 0 | 78 | 67% |
| GPT 5.4 Mini | Mid | 13 | 95 | 2 | 84 | 12% |
| DeepSeek R1 0528 | Frontier | 38 | 66 | 31 | 62 | 37% |
| Qwen3 Coder Next | Mid | 102 | 0 | 0 | 95 | 100% |
| GPT 5.4 | Frontier | 6 | 93 | 0 | 96 | 6% |
| Gemini 2.5 Flash | Small | 83 | 3 | 0 | 112 | 97% |
| Claude Sonnet 4.6 | Frontier | 0 | 77 | 0 | 121 | 0% |
| Gemini 2.5 Pro | Frontier | 0 | 73 | 0 | 125 | 0% |
| MiniMax M2.7 | Frontier | 45 | 26 | 0 | 125 | 63% |
| Kimi K2.5 | Frontier | 0 | 70 | 5 | 121 | 0% |
| Llama 4 Maverick | Frontier | 0 | 70 | 0 | 128 | 0% |
| Claude Haiku 4.5 | Small | 67 | 0 | 0 | 125 | 100% |
| GLM 5 Turbo | Frontier | 0 | 65 | 3 | 130 | 0% |
| MiMo V2 Pro | Frontier | 54 | 9 | 0 | 134 | 86% |
| Claude Opus 4.6 | Frontier | 0 | 49 | 0 | 148 | 0% |
| DeepSeek V3.2 | Mid | 2 | 12 | 1 | 183 | 14% |
| Prompt | Tier | Firebase | Ably | None | Other | A rate |
|---|---|---|---|---|---|---|
| chat-application | Advanced | 38 | 253 | 1 | 122 | 13% |
| chat-application | Intermediate | 155 | 84 | 6 | 171 | 65% |
| social-media-platform | Advanced | 25 | 187 | 3 | 202 | 12% |
| project-management-tool | Advanced | 30 | 156 | 0 | 231 | 16% |
| social-media-platform | Intermediate | 80 | 95 | 4 | 236 | 46% |
| chat-application | Beginner | 127 | 31 | 3 | 254 | 80% |
| project-management-tool | Intermediate | 100 | 44 | 3 | 268 | 69% |
| project-management-tool | Beginner | 120 | 18 | 5 | 274 | 87% |
| social-media-platform | Beginner | 69 | 61 | 17 | 263 | 53% |
