K2 Think API

K2's K2 Think, through an OpenAI-compatible API. Providers quote it on an open exchange, and each request goes to the offer with the lowest estimated charge that meets your rules.

Providers compete for each request. The exchange ranks the offers that meet your rules by estimated charge. It fixes your maximum charge before the first token, and you pay for the tokens that the model uses.
The OpenAI and Anthropic APIs. Use the OpenAI or Anthropic SDK that you have. Set the base URL and your key, and send k2-think as the model.
Your data rules. Send x-liquid-retention: none and only offers whose provider does not keep your content compete for the request.

Facts

Model name: k2-think
Maker: K2
Context window: 128K tokens
Input price: $0.170 per million tokens
Output price: $0.681 per million tokens
Providers: 1 independent provider
Regions: us

Providers

Architect: us, $0.170 input and $0.681 output per million tokens, $0.170 cache read per million tokens, 128K tokens of context

Scores

HLE: 10.1, #161 of 273
GPQA: 71.3, #148 of 250
AA-LCR: 57.0, #148 of 237
Terminal-Bench Hard: 6.8, #131 of 166
τ²-Bench: 25.4, #142 of 169
IFBench: 62.8, #63 of 178

Scores: Artificial Analysis (https://artificialanalysis.ai/)

Request

curl https://router.inference.ai.exchange/v1/chat/completions \
  -H "Authorization: Bearer $LIQUID_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "k2-think",
    "messages": [{"role":"user","content":"Say hello."}]
  }'

Questions

What is K2 Think?: K2 Think is a model from K2.
How much does K2 Think cost?: Offers now ask $0.170 per million input tokens and $0.681 per million output tokens. A job pays for the tokens it uses. The exchange picks the cheapest eligible offer.
What is the context length of K2 Think?: The largest offer takes 128,000 input tokens. A smaller offer can take less. The exchange skips an offer that cannot take your message.
What can K2 Think do?: Its providers list Reasoning effort, JSON object, JSON schema, Frequency penalty, Logit bias, Log probabilities, Min-p sampling, Presence penalty, Repetition penalty, Seed, Stop sequences, Top-k sampling, Top-p sampling, Text, Tools.
What inputs and outputs does K2 Think support?: Text → Text.
Which providers serve K2 Think?: 1 independent provider quote it now, with 1 offer. The table above ranks them by the cost of a typical job.
How do I call K2 Think?: Send "k2-think" as the model in an OpenAI or Anthropic request. The API section links to setup steps for each client.
When was K2 Think listed?: The exchange first listed it on 10/10/2026. This is not its release date.