GLM 4.6: providers compared
Every endpoint serving this model, side by side. ◆ marks the best value in each column — per model, never fleet-wide.
Current state
| State | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Healthy | 100%◆ (best) | 99.35% | 98.88% | 98.88% | 1.18s | 21.0s | 15 tps | 198K | $0.430 / $1.75◆ (best) | |
| Healthy | 99.42% | 99.91% | 99.27%◆ (best) | 99.27%◆ (best) | 751ms◆ (best) | 9.09s | 18 tps | 203K | $0.500 / $2.00 | |
| Healthy | 99.15% | 97.49% | 96.85% | 96.85% | 1.12s | 18.1s | 39 tps | 205K | $0.550 / $2.20 | |
| Healthy | 97.06% | 96.22% | 92.00% | 92.00% | 780ms | 14.9s | 38 tps | 203K | $0.600 / $2.20 | |
| Healthy | 95.35% | 93.49% | 96.87% | 96.87% | 818ms | 3.85s◆ (best) | 40 tps◆ (best) | 203K | $0.600 / $2.20 |
- Venice100%◆ (best)Healthy24h 99.35%
- 7d
- 98.88%
- 30d
- 98.88%
- p50
- 1.18s
- p99
- 21.0s
- Throughput
- 15 tps
- Context
- 198K
- $/Mtok
- $0.430 / $1.75◆ (best)
venice/fp4fp4 - DeepInfra99.42%Healthy24h 99.91%
- 7d
- 99.27%◆ (best)
- 30d
- 99.27%◆ (best)
- p50
- 751ms◆ (best)
- p99
- 9.09s
- Throughput
- 18 tps
- Context
- 203K
- $/Mtok
- $0.500 / $2.00
deepinfra/fp4fp4 - Novita99.15%Healthy24h 97.49%
- 7d
- 96.85%
- 30d
- 96.85%
- p50
- 1.12s
- p99
- 18.1s
- Throughput
- 39 tps
- Context
- 205K
- $/Mtok
- $0.550 / $2.20
novita/bf16bf16 - Z.AI97.06%Healthy24h 96.22%
- 7d
- 92.00%
- 30d
- 92.00%
- p50
- 780ms
- p99
- 14.9s
- Throughput
- 38 tps
- Context
- 203K
- $/Mtok
- $0.600 / $2.20
z-ai/fp4fp4 - AtlasCloud95.35%Healthy24h 93.49%
- 7d
- 96.87%
- 30d
- 96.87%
- p50
- 818ms
- p99
- 3.85s◆ (best)
- Throughput
- 40 tps◆ (best)
- Context
- 203K
- $/Mtok
- $0.600 / $2.20
atlas-cloud/fp8fp8
Measured 3m ago. Latency and throughput are OpenRouter's rolling 30-minute windows. ◆ marks the best value in each column among these endpoints.
Price vs latency
One dot per endpoint at the latest round · further left = faster, lower = cheaper · hover a dot for its tag
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, averaged over the hour.
Throughput
Median tokens per second.