Kimi K3 vs @cf/qwen/qwen2.5-coder-32b-instruct

Practical read: Kimi K3 reads more context. @cf/qwen/qwen2.5-coder-32b-instruct costs about a seventh as much. One API key runs both.

No card required. Start with 1 million free tokens.

Side by side:

MoonshotAI: Kimi K3 compared with @cf/qwen/qwen2.5-coder-32b-instruct
Attribute	Kimi K3 `moonshotai/kimi-k3`	@cf/qwen/qwen2.5-coder-32b-instruct `qwen/qwen2.5-coder-32b-instruct`
Price, 1M in + out	$11.97 Chutes, cheapest route	$1.75 Cloudflare Workers AI, cheapest route
Best for	Long documents	Low cost
Input / output, 1M	$1.95 / $10.02	$0.70 / $1.06
Cached input	$0.1688/1M to $0.910887/1M	Not published
Context window	1.05M 1,048,576 tokens	33K 32,768 tokens
First token, p50	0.82 s via Telnyx	Measuring
Throughput	255 tok/s via Wafer	Measuring
AI IQ	IQ 127 #22 on AI IQ	Unrated
Provider routes	26 with failover	1
Privacy	Confidential E2EE and zero retention route	Varies by provider
Recent route uptime range	66.67% to 100.00%	not enough data
Publisher	Kimi	Alibaba Cloud Model Studio
Modes	chat	chat
More detail	Overview Pricing Providers	Overview

Price adds one million input and one million output tokens on the cheapest route. Prices and speed change: see the monthly reports and run a small eval before you choose a default.

Test both with your agent

Paste this into Claude Code, Cursor or any coding agent.

Compare Kimi K3 and @cf/qwen/qwen2.5-coder-32b-instruct in this project through TrustedRouter.

1. Read https://trustedrouter.com/docs/llms.txt
2. Point the OpenAI-compatible client at
   https://api.trustedrouter.com/v1 and read the key from
   TRUSTEDROUTER_API_KEY.
3. Run our eval prompts on moonshotai/kimi-k3, then on qwen/qwen2.5-coder-32b-instruct.
4. Show me cost, latency and quality side by side before changing the default.

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.trustedrouter.com/v1",
    api_key=os.environ["TRUSTEDROUTER_API_KEY"],
)
for model in ("moonshotai/kimi-k3", "qwen/qwen2.5-coder-32b-instruct"):
    response = client.chat.completions.create(
        model=model,
        messages=[{"role": "user", "content": "What is the capital of France?"}],
    )
    print(model, response.choices[0].message.content)

for model in moonshotai/kimi-k3 qwen/qwen2.5-coder-32b-instruct; do
  curl https://api.trustedrouter.com/v1/chat/completions \
    -H "Authorization: Bearer $TRUSTEDROUTER_API_KEY" \
    -H "Content-Type: application/json" \
    -d "{\"model\": \"$model\", \"messages\": [{\"role\": \"user\", \"content\": \"What is the capital of France?\"}]}"
done

Your agent will need an API key.

Questions

Which should I use, MoonshotAI: Kimi K3 or @cf/qwen/qwen2.5-coder-32b-instruct?

MoonshotAI: Kimi K3 has more Credits provider routes and the larger context window. @cf/qwen/qwen2.5-coder-32b-instruct has the lower published input-plus-output rate. There is not enough recent probe data to compare speed.

Is MoonshotAI: Kimi K3 or @cf/qwen/qwen2.5-coder-32b-instruct cheaper?

The current cheapest TrustedRouter route is $11.97425/1M for MoonshotAI: Kimi K3 and $1.7513/1M for @cf/qwen/qwen2.5-coder-32b-instruct. The comparison uses current catalog prices and updates as provider pricing changes.

Is MoonshotAI: Kimi K3 or @cf/qwen/qwen2.5-coder-32b-instruct faster?

Current measured p50 time to first token is 823 ms for MoonshotAI: Kimi K3 and not enough data for @cf/qwen/qwen2.5-coder-32b-instruct. These are routed probe measurements, not vendor-advertised speeds, and update as new samples arrive.

Can I test MoonshotAI: Kimi K3 and @cf/qwen/qwen2.5-coder-32b-instruct with the same API?

Yes. Use the same OpenAI-compatible TrustedRouter base URL and API key, then change only the model id between moonshotai/kimi-k3 and qwen/qwen2.5-coder-32b-instruct. This makes side-by-side evals possible without maintaining two provider integrations.

More comparisons

All comparisons

Try Kimi K3 and @cf/qwen/qwen2.5-coder-32b-instruct with one key.

TrustedRouter is useful when you know the model you want, but still need provider rollover, budget limits, usage records, and a prompt path you can verify.

No card required. Start with 1 million free tokens.