Skip to content
xrouters
catalog

Every model, one price list.

The multiplier next to each id is what it costs against your quota. A 1× model burns one token per token used; a 2× model burns two. Nothing else changes between providers — same key, same endpoint, same response shape.

ModelPricingContext
Abliteration FP81 model
ablit-fp8-70b
0.25×64K
Claude2 models
claude-haiku-4.5
1×200K
claude-sonnet-4.5
3×200K
DeepSeek3 models
deepseek-coder-v3
0.6×128K
deepseek-v3.2
0.5×128K
deepseek-r1
1.5×128K
Gemini2 models
gemini-2.5-flash
0.5×1M
gemini-2.5-pro
2×1M
GLM (Zhipu)2 models
glm-4.5-air
0.4×128K
glm-4.6
1×200K
Kimi (Moonshot)2 models
kimi-k2-turbo
1.8×256K
kimi-k2
1.2×256K
MiMo (Xiaomi)1 model
mimo-7b-rl
0.3×128K
OpenAI1 model
gpt-5.1
2.5×400K
Qwen1 model
qwen3-max
1.4×256K

multiplier = tokens of quota spent per token used · latency measured p50 over 24h

Streaming by default

Server-sent events pass through unbuffered, including Anthropic's message deltas.

Usage from the source

Token counts come from the provider's usage block, not an estimate we made up.

Drop-in base URL

Swap the base URL in any OpenAI or Anthropic SDK — no wrapper, no shim, no new types.

Not sure which id to use?

Start with a balanced 1× model. Move to a 2× model only when the task needs the extra reasoning budget — the multiplier is the whole cost model.