Muse Spark vs GPT-5.6 Sol on HealthBench Hard
updated August 16, 2026
A gap of 0.097 separates these two on the 1,000 hardest HealthBench conversations: Muse Spark at 0.428, GPT-5.6 Sol at 0.331. The table adds what each one costs at list rates.
Side by side
 Muse Spark |  GPT-5.6 Sol | |
|---|---|---|
| score | 0.428 | 0.331 |
| rank | 1 of 9 | 2 of 9 |
| context window | 1.0M | 1.1M |
| price per 1M tokens, in / out | $1.25 / $4.25 | $5.00 / $30.00 |
| 1,000-exchange workload | $5.48 | $31.00 |
| released | 2026-08-05 | 2026-07-09 |
| license | proprietary | proprietary |
The workload row prices 1,000 exchanges of 2,000 input and 700 output tokens each, at the list rates current on August 16, 2026. GPT-5.6 models charge higher rates above 272K input tokens. GPT OSS models are open weights without vendor list pricing.
Reading the matchup
The cross-lab flagship pairing, and it is not close: 0.097 separates Muse Spark from the best GPT-5.6 model, several times any gap inside that tier. The unusual part is the price column. Muse Spark lists at $1.25 per million input tokens against Sol's $5.00, so on this benchmark Meta's model is both the stronger and the cheaper flagship. The case for Sol is the ecosystem and the 1.1M context window, not the score.
Which scores higher on HealthBench Hard, Muse Spark or GPT-5.6 Sol?
Muse Spark. On the 1,000-conversation set it scores 0.428 to GPT-5.6 Sol's 0.331, a margin of 0.097 as of August 16, 2026.
Which is cheaper to run, Muse Spark or GPT-5.6 Sol?
Muse Spark. The same workload of 1,000 exchanges (2,000 input and 700 output tokens each) comes to $5.48 on it and $31.00 on the other.
Related comparisons
- 0.331 vs 0.327
- 0.331 vs 0.016
Each model's full page: Muse Spark and GPT-5.6 Sol. Every other pairing lives in the score-difference matrix.