Which model fits your task

Prices from our price list. Speed and quality are indicative estimates, not live measurements.

We recommend

Kimi K2.6

Strong at long IDE tasks and careful with tools.

Per answer
0.26 pts
Speed
55 tok/s
Quality
86
Cheapest
0.0042 pts / answer

Llama 3.1 8B

Fastest
220 tok/s

Llama 3.1 8B

Smartest
86 / 100

Kimi K2.6

First token
0.3 s

Llama 3.1 8B

Answer price

Lower is better

  • Llama 3.1 8B
    0.0042 pts
  • gpt-oss-120b
    0.013 pts
  • Phi-4
    0.015 pts
  • Mistral Small 3.2
    0.018 pts
  • Llama 4 Scout
    0.027 pts
  • Llama 3.3 70B
    0.028 pts
  • Qwen3 235B
    0.039 pts
  • DeepSeek V3.2
    0.048 pts
  • DeepSeek V3.1
    0.077 pts
  • Qwen3 Coder
    0.085 pts
  • GLM-4.7
    0.13 pts
  • DeepSeek R1
    0.17 pts
  • Kimi K2.6
    0.26 pts

Price ↔ quality

Top-left corner is the best deal

8085909500.091 pts0.18 pts0.27 ptsKimi K2.6GLM-4.7DeepSeek R1DeepSeek V3.1DeepSeek V3.2Qwen3 235BQwen3 CoderLlama 4 ScoutLlama 3.3 70BMistral Small 3.2gpt-oss-120bPhi-4Llama 3.1 8B

Time to first token, 14 days

Median across all requests, seconds

00.81.62.4Sep 11Sep 13Sep 15Sep 17Sep 19Sep 21Sep 23

Head to head

Pick two models — the winner on each metric is highlighted

Price per answer0.13 pts0.077 pts ✓
Input / 1M42 pts27 pts ✓
Output / 1M184 pts100 pts ✓
Speed60 tok/s ✓58 tok/s
First token1.1 s ✓1.6 s
Quality85 ✓84
Context200K160K