Compare models
updated September 8, 2026
Matchups that decide deployments get a full page each, with the score gap, list prices, and a costed usage workload. The matrix below covers every remaining pairing.
Head-to-head pages
- 0.428 vs 0.331 · Muse Spark by 0.097
- 0.331 vs 0.327 · GPT-5.6 Sol by 0.004
- 0.320 vs 0.300 · GPT-5.6 Luna by 0.020
- 0.300 vs 0.108 · GPT OSS 120B by 0.192
- 0.347 vs 0.331 · GPT-5 by 0.016
- 0.320 vs 0.259 · GPT-5.6 Luna by 0.061
Score-difference matrix
Each cell is the row model's score minus the column model's, so positive means the row wins. Column numbers follow rank.
| model | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Muse Spark | · | +0.065 | +0.081 | +0.085 | +0.097 | +0.101 | +0.108 | +0.113 | +0.114 | +0.128 | +0.137 | +0.141 | +0.169 | +0.174 | +0.199 | +0.320 |
| 2. GPT-6 Astra | -0.065 | · | +0.016 | +0.020 | +0.032 | +0.036 | +0.043 | +0.048 | +0.049 | +0.063 | +0.072 | +0.076 | +0.104 | +0.109 | +0.134 | +0.255 |
| 3. GPT-5 | -0.081 | -0.016 | · | +0.004 | +0.016 | +0.020 | +0.027 | +0.032 | +0.033 | +0.047 | +0.056 | +0.060 | +0.088 | +0.093 | +0.118 | +0.239 |
| 4. GPT-5.2 | -0.085 | -0.020 | -0.004 | · | +0.012 | +0.016 | +0.023 | +0.028 | +0.029 | +0.043 | +0.052 | +0.056 | +0.084 | +0.089 | +0.114 | +0.235 |
| 5. GPT-5.6 Sol | -0.097 | -0.032 | -0.016 | -0.012 | · | +0.004 | +0.011 | +0.016 | +0.017 | +0.031 | +0.040 | +0.044 | +0.072 | +0.077 | +0.102 | +0.223 |
| 6. GPT-5.6 Terra | -0.101 | -0.036 | -0.020 | -0.016 | -0.004 | · | +0.007 | +0.012 | +0.013 | +0.027 | +0.036 | +0.040 | +0.068 | +0.073 | +0.098 | +0.219 |
| 7. GPT-5.6 Luna | -0.108 | -0.043 | -0.027 | -0.023 | -0.011 | -0.007 | · | +0.005 | +0.006 | +0.020 | +0.029 | +0.033 | +0.061 | +0.066 | +0.091 | +0.212 |
| 8. GPT-5.5 | -0.113 | -0.048 | -0.032 | -0.028 | -0.016 | -0.012 | -0.005 | · | +0.001 | +0.015 | +0.024 | +0.028 | +0.056 | +0.061 | +0.086 | +0.207 |
| 9. GPT-5.6 Sol (August) | -0.114 | -0.049 | -0.033 | -0.029 | -0.017 | -0.013 | -0.006 | -0.001 | · | +0.014 | +0.023 | +0.027 | +0.055 | +0.060 | +0.085 | +0.206 |
| 10. GPT OSS 120B | -0.128 | -0.063 | -0.047 | -0.043 | -0.031 | -0.027 | -0.020 | -0.015 | -0.014 | · | +0.009 | +0.013 | +0.041 | +0.046 | +0.071 | +0.192 |
| 11. GPT-5.4 | -0.137 | -0.072 | -0.056 | -0.052 | -0.040 | -0.036 | -0.029 | -0.024 | -0.023 | -0.009 | · | +0.004 | +0.032 | +0.037 | +0.062 | +0.183 |
| 12. GPT-5.6 Luna (August) | -0.141 | -0.076 | -0.060 | -0.056 | -0.044 | -0.040 | -0.033 | -0.028 | -0.027 | -0.013 | -0.004 | · | +0.028 | +0.033 | +0.058 | +0.179 |
| 13. GPT-5.3 Chat | -0.169 | -0.104 | -0.088 | -0.084 | -0.072 | -0.068 | -0.061 | -0.056 | -0.055 | -0.041 | -0.032 | -0.028 | · | +0.005 | +0.030 | +0.151 |
| 14. GPT-5.1 | -0.174 | -0.109 | -0.093 | -0.089 | -0.077 | -0.073 | -0.066 | -0.061 | -0.060 | -0.046 | -0.037 | -0.033 | -0.005 | · | +0.025 | +0.146 |
| 15. GPT-5.5 Instant | -0.199 | -0.134 | -0.118 | -0.114 | -0.102 | -0.098 | -0.091 | -0.086 | -0.085 | -0.071 | -0.062 | -0.058 | -0.030 | -0.025 | · | +0.121 |
| 16. GPT OSS 20B | -0.320 | -0.255 | -0.239 | -0.235 | -0.223 | -0.219 | -0.212 | -0.207 | -0.206 | -0.192 | -0.183 | -0.179 | -0.151 | -0.146 | -0.121 | · |
On a 1,000-conversation set, anything within 0.01 sits inside the variation between two measurements of the same model; read those cells as ties. Rows read from different documents carry the extra caveats on the methodology page; the document behind each row is on the sources page.
Pricing and context windows for every model are on the models page.