Claude Haiku 4.5: 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 | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
Anthropicanthropic | Healthy | 100%◆ (best) | 99.94% | 99.90% | 99.94% | 655ms | 2.10s | 78 tps | 200K | $1.00 / $5.00◆ (best) |
Googlegoogle-vertex/global | Healthy | 100%◆ (best) | 99.39% | 99.63% | 99.74% | 729ms | 4.20s | 84 tps | 200K | $1.00 / $5.00◆ (best) |
Amazon Bedrockamazon-bedrock/global | Healthy | 99.98% | 99.95% | 99.95% | 99.96% | 772ms | 3.49s | 73 tps | 200K | $1.00 / $5.00◆ (best) |
Amazon Bedrockamazon-bedrock/eu-west-1 | No data | — | 99.96% | 99.92% | 99.96% | 5.93s | 15.9s | 128 tps | 200K | $1.10 / $5.50 |
Amazon Bedrockamazon-bedrock/us | No data | — | 99.91% | 99.74% | 99.83% | 1.12s | 2.02s | 30 tps | 200K | $1.10 / $5.50 |
Azureazure/global | No data | — | 99.99% | 99.74% | 99.89% | 837ms | 1.91s | 65 tps | 200K | $1.00 / $5.00◆ (best) |
Googlegoogle-vertex/europe | No data | — | 100% | 100% | 99.99%◆ (best) | 500ms◆ (best) | 1.23s◆ (best) | 81 tps | 200K | $1.10 / $5.50 |
Googlegoogle-vertex/us-east5 | No data | — | 99.94% | 100%◆ (best) | 99.96% | 2.58s | 2.98s | 194 tps◆ (best) | 200K | $1.10 / $5.50 |
- Anthropic100%◆ (best)Healthy24h 99.94%
- 7d
- 99.90%
- 30d
- 99.94%
- p50
- 655ms
- p99
- 2.10s
- Throughput
- 78 tps
- Context
- 200K
- $/Mtok
- $1.00 / $5.00◆ (best)
anthropic - Google100%◆ (best)Healthy24h 99.39%
- 7d
- 99.63%
- 30d
- 99.74%
- p50
- 729ms
- p99
- 4.20s
- Throughput
- 84 tps
- Context
- 200K
- $/Mtok
- $1.00 / $5.00◆ (best)
google-vertex/global - Amazon Bedrock99.98%Healthy24h 99.95%
- 7d
- 99.95%
- 30d
- 99.96%
- p50
- 772ms
- p99
- 3.49s
- Throughput
- 73 tps
- Context
- 200K
- $/Mtok
- $1.00 / $5.00◆ (best)
amazon-bedrock/global - No data24h 99.96%
- 7d
- 99.92%
- 30d
- 99.96%
- p50
- 5.93s
- p99
- 15.9s
- Throughput
- 128 tps
- Context
- 200K
- $/Mtok
- $1.10 / $5.50
amazon-bedrock/eu-west-1 - No data24h 99.91%
- 7d
- 99.74%
- 30d
- 99.83%
- p50
- 1.12s
- p99
- 2.02s
- Throughput
- 30 tps
- Context
- 200K
- $/Mtok
- $1.10 / $5.50
amazon-bedrock/us - No data24h 99.99%
- 7d
- 99.74%
- 30d
- 99.89%
- p50
- 837ms
- p99
- 1.91s
- Throughput
- 65 tps
- Context
- 200K
- $/Mtok
- $1.00 / $5.00◆ (best)
azure/global - No data24h 100%
- 7d
- 100%
- 30d
- 99.99%◆ (best)
- p50
- 500ms◆ (best)
- p99
- 1.23s◆ (best)
- Throughput
- 81 tps
- Context
- 200K
- $/Mtok
- $1.10 / $5.50
google-vertex/europe - No data24h 99.94%
- 7d
- 100%◆ (best)
- 30d
- 99.96%
- p50
- 2.58s
- p99
- 2.98s
- Throughput
- 194 tps◆ (best)
- Context
- 200K
- $/Mtok
- $1.10 / $5.50
google-vertex/us-east5
Measured 1m 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.
Throughput
Median tokens per second.