EQ-Bench 4 leaderboard
- Claude Opus 5 leads
- 1385
- models measured
- 28
- lead over #2
- +45 Elo
- median
- 1215
Leader on EQ-Bench 4 · read October 22
On October 22, 2026, 28 of our catalog's models had a EQ-Bench 4 result, and Anthropic Claude Opus 5 led with 1385, ahead of Anthropic Claude Fable 5 at 1340.
EQ-Bench 4 tests emotional intelligence. A model answers in role-played conversations, and judge models compare the answers pairwise. The number is the resulting Elo rating, as published on eqbench.com.
Role in the index
EQ-Bench 4 is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Interval (±) | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|---|
| 1 | Claude Opus 5Anthropic | 1385 | ±24.8 | 83.0 | $25.00 | |
| 2 | Claude Fable 5Anthropic | 1340 | ±13.4 | 79.3 | $50.00 | |
| 3 | Kimi K3Moonshot AI | 1339 | ±15.3 | 62.3 | $15.00 | |
| 4 | GPT-5.5OpenAI | 1315 | ±11.9 | 71.3 | $30.00 | |
| 5 | Claude Opus 4.7Anthropic | 1311 | ±11.3 | 67.5 | $25.00 | |
| 6 | Claude Opus 4.8Anthropic | 1281 | ±11.1 | 65.0 | $25.00 | |
| 7 | GPT-5.4OpenAI | 1272 | ±10.0 | 64.4 | $15.00 | |
| 8 | Muse Spark 1.1Meta | 1260 | ±13.5 | 70.3 | $4.25 | |
| 9 | GPT-5.6 SolOpenAI | 1250 | ±11.8 | 73.5 | $20.00 | |
| 10 | Claude Sonnet 5Anthropic | 1236 | ±11.3 | 57.7 | $10.00 | |
| 11 | GPT-5.6 TerraOpenAI | 1234 | ±9.6 | 63.9 | $12.00 | |
| 12 | T InklingThinking Machines | 1226 | ±12.0 | 31.5 | $4.05 | |
| 13 | Claude Opus 4.6Anthropic | 1223 | ±8.8 | 57.2 | $25.00 | |
| 14 | GLM 5.2Z.ai | 1222 | ±9.4 | 61.1 | $4.40 | |
| 15 | MiMo-V2.5-ProXiaomi | 1208 | ±10.7 | 31.6 | $0.9152 | |
| 16 | Claude Sonnet 4.6Anthropic | 1207 | ±8.8 | 57.3 | $15.00 | |
| 17 | Kimi K2.6Moonshot AI | 1202 | ±8.6 | 52.1 | $4.00 | |
| 18 | DeepSeek V4 Pro 0423DeepSeek | 1166 | ±9.0 | 38.7 | $2.60 | |
| 19 | GPT-5.6 LunaOpenAI | 1156 | ±12.8 | 58.0 | $1.20 | |
| 20 | MiniMax M3MiniMax | 1150 | ±13.1 | 39.2 | $1.20 | |
| 21 | Gemini 3.1 Pro PreviewGoogle | 1142 | ±12.0 | 32.3 | $12.00 | |
| 22 | Gemma 4 31BGoogle | 1120 | ±12.4 | 39.0 | $0.40 | |
| 23 | Qwen3.7 MaxQwen | 1110 | ±11.9 | 54.1 | $4.425 | |
| 24 | Gemini 3.5 FlashGoogle | 1087 | ±12.1 | 39.8 | $9.00 | |
| 25 | Grok 4.3xAI | 1075 | ±15.9 | 37.4 | $2.50 | |
| 26 | Claude Haiku 4.5Anthropic | 1064 | ±11.9 | 35.6 | $5.00 | |
| 27 | Qwen3.6 27BQwen | 1026 | ±20.3 | 36.8 | $2.70 | |
| 28 | Mistral Medium 3.5Mistral AI | 993 | ±18.6 | 25.5 | $7.50 |
Only models in our catalog are listed, and each appears once with its best published run. 28 rows were read from the source on October 22, 2026, and rows that match no catalog model are left out. Every other column is a figure the board publishes for the same run.