StreamLake — pricing & performance
Models served
20
of 908 tracked models
Mean output speed
51
tokens per second, averaged over
20 measured endpoints
20 measured endpoints
Mean latency
1.5s
time to first token, averaged over
20 measured endpoints
20 measured endpoints
Cheapest blended
$0.084
per 1M tokens at a 3:1
input:output blend
input:output blend
Output speed on StreamLake
Median output tokens per second for each model this host serves, derived from OpenRouter endpoint telemetry. Higher is better. We do not run these measurements ourselves.
12 of 20 models ▾
☰⚙⇩
+Add model from specific provider
The Known Good
20 models on StreamLake
Full leaderboard →
| Model | Creator | Input $/M | Output $/M | Blended | Speed | TTFT | Context |
|---|---|---|---|---|---|---|---|
DeepSeek: DeepSeek V3
|
DeepSeek | $0.20 | $0.80 | $0.35 | 24 | 1.3s | 128k |
|
DeepSeek: DeepSeek V3.1 Terminus
|
DeepSeek | $0.34 | $1.03 | $0.51 | 30 | 1.2s | 128k |
|
DeepSeek: DeepSeek V3.2
|
DeepSeek | $0.21 | $0.32 | $0.24 | 25 | 1.4s | 128k |
|
DeepSeek: DeepSeek V4 Flash
|
DeepSeek | $0.092 | $0.18 | $0.12 | 51 | 1.7s | 1.02M |
|
DeepSeek: DeepSeek V4 Pro
|
DeepSeek | $0.67 | $1.34 | $0.84 | 41 | 1.6s | 1.02M |
|
DeepSeek: R1 0528
|
DeepSeek | $0.57 | $2.29 | $1.00 | 32 | 2.6s | 128k |
Kwaipilot: KAT-Coder-Air V2.5
|
Kuaishou | $0.15 | $0.60 | $0.26 | 101 | 1.8s | 256k |
|
Kwaipilot: KAT-Coder-Pro V2
|
Kuaishou | $0.30 | $1.20 | $0.53 | 66 | 1.1s | 256k |
|
Kwaipilot: KAT-Coder-Pro V2.5
|
Kuaishou | $0.74 | $2.96 | $1.29 | 55 | 1.0s | 256k |
MiniMax: MiniMax M2.5
|
MiniMax | $0.27 | $1.08 | $0.47 | 72 | 0.8s | 200k |
MoonshotAI: Kimi K2.5
|
Moonshot AI | $0.54 | $2.70 | $1.08 | 55 | 1.7s | 256k |
|
MoonshotAI: Kimi K2.6
|
Moonshot AI | $0.85 | $3.60 | $1.54 | 44 | 1.6s | 256k |
Qwen: Qwen3 235B A22B Instruct 2507
|
Alibaba | $0.21 | $0.84 | $0.37 | 37 | 0.6s | 128k |
|
Qwen: Qwen3 30B A3B Instruct 2507
|
Alibaba | $0.048 | $0.19 | $0.084 | 42 | 0.9s | 128k |
|
Qwen: Qwen3 Coder Next
|
Alibaba | $0.18 | $0.90 | $0.36 | 50 | 1.4s | 256k |
|
Qwen: Qwen3.5 397B A17B
|
Alibaba | $0.60 | $3.60 | $1.35 | 140 | 0.8s | 256k |
Z.ai: GLM 4.7
|
Z.ai | $0.48 | $1.76 | $0.80 | 41 | 3.0s | 200k |
|
Z.ai: GLM 5
|
Z.ai | $0.60 | $1.92 | $0.93 | 35 | 3.1s | 198k |
|
Z.ai: GLM 5.1
|
Z.ai | $0.97 | $3.04 | $1.48 | 25 | 1.4s | 200k |
|
Z.ai: GLM 5.2
|
Z.ai | $0.70 | $2.20 | $1.07 | 45 | 1.6s | 1.02M |
The Known Good
Source. Prices, endpoint availability and throughput for StreamLake are ingested from OpenRouter;
model metadata is enriched from models.dev. A blank cell means the figure was not published for that endpoint —
it never means zero. Blended price is (3 × input + output) ÷ 4, the same 3:1 blend used everywhere on this
site; see methodology. Quality scores are not provider-specific:
a model scores the same wherever it is hosted, so quality lives on the
model pages.