Best-value LLM APIs (speed per dollar)

As of 2026-06-21, the top model is deepseek-v4-flash (142 tok/s @ $0.28/1M). See the full ranking below — click a model for details or run your own test.

⚡ Fastest💰 Cheapest⭐ Best value
  1. 🥇
    deepseek-v4-flashapi.deepseek.com
    top speed-per-dollar
    142 tok/s @ $0.28/1M
  2. 🥈
    mimo-v2.5api.xiaomimimo.com
    top speed-per-dollar
    85 tok/s @ $0.28/1M
  3. 🥉
    step-3.7-flashapi.stepfun.com
    top speed-per-dollar
    162 tok/s @ $1.1/1M
  4. 4
    mimo-v2.5-proapi.xiaomimimo.com
    top speed-per-dollar
    37 tok/s @ $0.28/1M
  5. 5
    deepseek-v4-proapi.deepseek.com
    top speed-per-dollar
    81 tok/s @ $0.87/1M
  6. 6
    kimi-for-codingapi.kimi.com
    top speed-per-dollar
    115 tok/s @ $4.0/1M
  7. 7
    glm-5.1open.bigmodel.cn
    top speed-per-dollar
    53 tok/s @ $4.4/1M
  8. 8
    glm-5.2open.bigmodel.cn
    top speed-per-dollar
    49 tok/s @ $4.4/1M
Full pricing →Speed leaderboard →Run your own test ▶
📋 Embed this ranking on your site

Copy the code below (live data, auto-updating):

<iframe src="https://www.tokrace.com/embed/best/value?lang=en" width="400" height="360" style="border:0;border-radius:12px" loading="lazy" title="Best-value LLM APIs (speed per dollar)"></iframe>

· Speed is the median of anonymous benchmarks; prices follow official pages and cross-currency uses a fixed USD rate. Reference only.