deepseek/deepseek-chat

Aliases: deepseek-v3, v3

$0.32

Input / 1M

$0.89

Output / 1M

164K

Context

50 t/s

Speed

Public benchmark scores

Sourced from each provider's published numbers. Verify before quoting.

Quality index

MMLU

87.1

GPQA

59.1

HumanEval

MATH

90.2

SWE-bench

—

Arena Elo

—

Sources: deepseek-v3-paper

Capabilities

json_modestreamingstructured_outputtools

Use deepseek-chat via Relay

Configure the model alias in YAML, then call it from Python.

YAML

# models.yaml
version: 1
models:
  deepseek-v3:
    target: deepseek/deepseek-chat
    credential: $env.DEEPSEEK_API_KEY

Python

from relay import Hub

async with Hub.from_yaml("models.yaml") as hub:
    resp = await hub.chat(
        "deepseek-v3",
        messages=[{"role": "user", "content": "Hello"}],
    )
    print(resp.text, resp.cost_usd)

pip install ai5labs-relay · full docs on GitHub

Compare with

vs xai/grok-3 vs anthropic/claude-sonnet-4-5 vs openai/gpt-4o vs anthropic/claude-3-5-sonnet-20241022