MiniMax M3
minimax/minimax-m3
Compare with MiniMax M2.7, MiniMax M2.5, MiniMax M3 (batch)
CompareMiniMax M3
| State | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Healthy | 100% | 99.68% | 97.24% | 97.24% | 519ms | 3.08s | 81 tps | 262K | $0.230 / $0.960 | |
| Healthy | 100% | 99.51% | 99.33% | 99.33% | 3.79s | 11.7s | 44 tps | 1.0M | $0.240 / $0.960 | |
| Healthy | 100% | 99.53% | 99.63% | 99.63% | 1.60s | 17.7s | 52 tps | 524K | $0.300 / $1.20 | |
| Healthy | 100% | 99.76% | 99.76% | 99.76% | 1.70s | 10.2s | 132 tps | 1.0M | $0.750 / $3.00 | |
| Healthy | 100% | 96.59% | 97.49% | 97.49% | 1.32s | 20.9s | 21 tps | 256K | $0.255 / $1.02 | |
SambaNovasambanova | Healthy | 100% | 97.23% | 98.38% | 98.38% | 2.49s | 10.6s | 164 tps | 1.0M | $0.600 / $2.40 |
| Healthy | 100% | 99.02% | 98.46% | 98.46% | 1.77s | 9.89s | 49 tps | 1.0M | $0.300 / $1.20 | |
Togethertogether | Healthy | 100% | 99.78% | 99.78% | 99.78% | 758ms | 4.95s | 76 tps | 524K | $0.300 / $1.20 |
| Healthy | 100% | 99.82% | 99.87% | 99.87% | 1.27s | 11.0s | 110 tps | 524K | $0.300 / $1.20 | |
| Healthy | 99.81% | 98.34% | 97.91% | 97.91% | 1.10s | 14.9s | 46 tps | 1.0M | $0.300 / $1.20 | |
| Healthy | 99.02% | 99.64% | 99.66% | 99.66% | 2.40s | 21.1s | 37 tps | 1.0M | $0.300 / $1.20 | |
| Healthy | 98.48% | 99.71% | 99.52% | 99.52% | 1.29s | 5.96s | 95 tps | 524K | $0.300 / $1.20 | |
| Degraded | 91.55% | 98.43% | 98.38% | 98.38% | 741ms | 9.71s | 31 tps | 524K | $0.280 / $1.10 |
- CoreWeave100%Healthy24h 99.68%
- 7d
- 97.24%
- 30d
- 97.24%
- p50
- 519ms
- p99
- 3.08s
- Throughput
- 81 tps
- Context
- 262K
- $/Mtok
- $0.230 / $0.960
coreweave/fp4fp4 - GMICloud100%Healthy24h 99.51%
- 7d
- 99.33%
- 30d
- 99.33%
- p50
- 3.79s
- p99
- 11.7s
- Throughput
- 44 tps
- Context
- 1.0M
- $/Mtok
- $0.240 / $0.960
gmicloud/fp8fp8 - Minimax100%Healthy24h 99.53%
- 7d
- 99.63%
- 30d
- 99.63%
- p50
- 1.60s
- p99
- 17.7s
- Throughput
- 52 tps
- Context
- 524K
- $/Mtok
- $0.300 / $1.20
minimax/fp8fp8 - ModelRun100%Healthy24h 99.76%
- 7d
- 99.76%
- 30d
- 99.76%
- p50
- 1.70s
- p99
- 10.2s
- Throughput
- 132 tps
- Context
- 1.0M
- $/Mtok
- $0.750 / $3.00
modelrun/fp4fp4 - Morph100%Healthy24h 96.59%
- 7d
- 97.49%
- 30d
- 97.49%
- p50
- 1.32s
- p99
- 20.9s
- Throughput
- 21 tps
- Context
- 256K
- $/Mtok
- $0.255 / $1.02
morph/fp4fp4 - SambaNova100%Healthy24h 97.23%
- 7d
- 98.38%
- 30d
- 98.38%
- p50
- 2.49s
- p99
- 10.6s
- Throughput
- 164 tps
- Context
- 1.0M
- $/Mtok
- $0.600 / $2.40
sambanova - StreamLake100%Healthy24h 99.02%
- 7d
- 98.46%
- 30d
- 98.46%
- p50
- 1.77s
- p99
- 9.89s
- Throughput
- 49 tps
- Context
- 1.0M
- $/Mtok
- $0.300 / $1.20
streamlake/fp8fp8 - Together100%Healthy24h 99.78%
- 7d
- 99.78%
- 30d
- 99.78%
- p50
- 758ms
- p99
- 4.95s
- Throughput
- 76 tps
- Context
- 524K
- $/Mtok
- $0.300 / $1.20
together - Venice100%Healthy24h 99.82%
- 7d
- 99.87%
- 30d
- 99.87%
- p50
- 1.27s
- p99
- 11.0s
- Throughput
- 110 tps
- Context
- 524K
- $/Mtok
- $0.300 / $1.20
venice/fp8fp8 - Parasail99.81%Healthy24h 98.34%
- 7d
- 97.91%
- 30d
- 97.91%
- p50
- 1.10s
- p99
- 14.9s
- Throughput
- 46 tps
- Context
- 1.0M
- $/Mtok
- $0.300 / $1.20
parasail/fp8fp8 - Novita99.02%Healthy24h 99.64%
- 7d
- 99.66%
- 30d
- 99.66%
- p50
- 2.40s
- p99
- 21.1s
- Throughput
- 37 tps
- Context
- 1.0M
- $/Mtok
- $0.300 / $1.20
novita/fp8fp8 - AtlasCloud98.48%Healthy24h 99.71%
- 7d
- 99.52%
- 30d
- 99.52%
- p50
- 1.29s
- p99
- 5.96s
- Throughput
- 95 tps
- Context
- 524K
- $/Mtok
- $0.300 / $1.20
atlas-cloud/fp8fp8 - DeepInfra91.55%Degraded24h 98.43%
- 7d
- 98.38%
- 30d
- 98.38%
- p50
- 741ms
- p99
- 9.71s
- Throughput
- 31 tps
- Context
- 524K
- $/Mtok
- $0.280 / $1.10
deepinfra/fp8fp8
Measured 3m ago. Latency and throughput are OpenRouter's rolling 30-minute windows.
Price changes
- MiniMax M3Morphmorph/fp4$0.300 / $1.20 → $0.255 / $1.022026-08-21 01:15 UTC
- MiniMax M3SambaNovasambanovafree / free → $0.600 / $2.402026-08-20 21:35 UTC
History
Availability
Axis starts at 90% so small differences stay visible; it expands to 0 when an endpoint falls below.
Time to first token
Logarithmic axis — the fleet spans two orders of magnitude. P50 is OpenRouter's rolling 30-minute window.
Throughput
Median tokens per second.