Meta vs 
Reflection
Verdict
- Meta's flagship leads: Muse Wonder Full at 68.6 vs Beam 1.0 at 56.4.
- Meta has the better model in 3 of 4 price bands.
- Meta's fastest ranked model (Muse Spark 1.2, 372 tok/s) beats Reflection's (Beam 1.0, 303 tok/s).
- Meta publishes more open-weights models (19 vs 2).
Best model per price band
Each provider’s highest-index model below the price ceiling. Purple marks the winner of the band.
| Price band | Meta | Reflection |
|---|---|---|
| Under $0.50/M | Llama 4 Maverick 42.9$0.418/M | – |
| Under $2/M | Muse Glimmer 54.9$0.525/M | Beam 1.0 56.4$1.05/M |
| Under $10/M | Muse Spark 1.3 max 67.2$2.00/M | Beam 1.0 56.4$1.05/M |
| Any price | Muse Spark 1.3 max 67.2$2.00/M | Beam 1.0 56.4$1.05/M |
Line-up
| Metric | Meta | Reflection |
|---|---|---|
| Flagship index | 68.6 | 56.4 |
| Ranked models | 27 | 1 |
| Open-weights models | 19 | 2 |
| Cheapest ranked, $/M | $0.048 | $1.05 |
| Fastest ranked, tok/s | 372 | 303 |
| Release cadence, days | 27 | – |
| Last release | 15 Oct 2026 | – |
| Next expected | 31 Oct 2026 – 21 Nov 2026 | – |
Data as of 18 Oct 2026. See Meta and Reflection in full.