modelstatus.dev

Kimi K2.5

moonshotai/kimi-k2.5

Compare with Kimi K2.6, Kimi K2.7 Code, Kimi K3

CompareKimi K2.5
  • DeepInfra100%
    Healthy24h 99.28%
    p50
    696ms
    p99
    7.06s
    Throughput
    43 tps
    Context
    262K
    $/Mtok
    $0.450 / $2.25
    deepinfra/fp4fp4
  • DigitalOcean100%
    Healthy24h 99.40%
    p50
    1.02s
    p99
    5.59s
    Throughput
    12 tps
    Context
    262K
    $/Mtok
    $0.375 / $2.02
    digitalocean
  • Moonshot AI100%
    Healthy24h 99.98%
    p50
    2.22s
    p99
    10.9s
    Throughput
    24 tps
    Context
    262K
    $/Mtok
    $0.600 / $3.00
    moonshotai/int4int4
  • SiliconFlow100%
    Healthy24h 99.91%
    p50
    1.50s
    p99
    38.1s
    Throughput
    39 tps
    Context
    262K
    $/Mtok
    $0.450 / $2.25
    siliconflow/int4int4
  • Venice100%
    Healthy24h 95.90%
    p50
    2.25s
    p99
    13.4s
    Throughput
    30 tps
    Context
    256K
    $/Mtok
    $0.532 / $3.32
    venice
  • Novita95.93%
    Healthy24h 98.38%
    p50
    855ms
    p99
    13.2s
    Throughput
    34 tps
    Context
    262K
    $/Mtok
    $0.570 / $2.85
    novita
  • AtlasCloud95.01%
    Degraded24h 98.43%
    p50
    1.55s
    p99
    10.5s
    Throughput
    28 tps
    Context
    262K
    $/Mtok
    $0.490 / $2.50
    atlas-cloud/int4int4
  • Amazon Bedrock
    No data24h 96.74%
    p50
    1.09s
    p99
    7.30s
    Throughput
    29 tps
    Context
    262K
    $/Mtok
    $0.600 / $3.00
    amazon-bedrock/us-east-2
  • Phala
    No data24h 83.86%
    p50
    2.49s
    p99
    128s
    Throughput
    32 tps
    Context
    262K
    $/Mtok
    $0.600 / $3.00
    phala
  • StreamLake
    No data24h 99.91%
    p50
    1.50s
    p99
    9.53s
    Throughput
    31 tps
    Context
    256K
    $/Mtok
    $0.540 / $2.70
    streamlake/fp8fp8

Measured 3m ago. Latency and throughput are OpenRouter's rolling 30-minute windows.

3 endpoints reporting no data are omitted from the charts.

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 and p99 are OpenRouter's rolling 30-minute windows.

Throughput

Median tokens per second.