gpt-oss-120b vs Qwen3 235B A22B Thinking 2507

Practical read: gpt-oss-120b costs under a tenth as much. One API key runs both.

No card required. Start with 1 million free tokens.

Side by side:

OpenAI: gpt-oss-120b compared with Qwen: Qwen3 235B A22B Thinking 2507
Attribute	gpt-oss-120b `openai/gpt-oss-120b`	Qwen3 235B A22B Thinking 2507 `qwen/qwen3-235b-a22b-thinking-2507`
Price, 1M in + out	$0.21 Weights & Biases, cheapest route	$4.17 Featherless AI, cheapest route
Best for	Low cost
Input / output, 1M	$0.032 / $0.18	$0.47 / $3.69
Cached input	$0.015825/1M to $0.1055/1M	Not published
Context window	131K 131,072 tokens	128K 128,000 tokens
First token, p50	0.48 s via Cerebras	Measuring
AI IQ	IQ 98 #136 on AI IQ	Unrated
Provider routes	24 with failover	2 with failover
Privacy	Confidential E2EE and zero retention route	Zero retention route
Recent route uptime range	100.00%	not enough data
Publisher	OpenAI	Alibaba Cloud Model Studio
Modes	chat	chat
More detail	Overview Pricing Providers	Overview Pricing Providers

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 gpt-oss-120b and Qwen3 235B A22B Thinking 2507 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 openai/gpt-oss-120b, then on qwen/qwen3-235b-a22b-thinking-2507.
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 ("openai/gpt-oss-120b", "qwen/qwen3-235b-a22b-thinking-2507"):
    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 openai/gpt-oss-120b qwen/qwen3-235b-a22b-thinking-2507; 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, OpenAI: gpt-oss-120b or Qwen: Qwen3 235B A22B Thinking 2507?

gpt-oss-120b has the lower published input-plus-output rate, more Credits provider routes and the larger context window. There is not enough recent probe data to compare speed.

Is OpenAI: gpt-oss-120b or Qwen: Qwen3 235B A22B Thinking 2507 cheaper?

The current cheapest TrustedRouter route is $0.211/1M for OpenAI: gpt-oss-120b and $4.16725/1M for Qwen: Qwen3 235B A22B Thinking 2507. The comparison uses current catalog prices and updates as provider pricing changes.

Is OpenAI: gpt-oss-120b or Qwen: Qwen3 235B A22B Thinking 2507 faster?

Current measured p50 time to first token is 482 ms for OpenAI: gpt-oss-120b and not enough data for Qwen: Qwen3 235B A22B Thinking 2507. These are routed probe measurements, not vendor-advertised speeds, and update as new samples arrive.

Can I test OpenAI: gpt-oss-120b and Qwen: Qwen3 235B A22B Thinking 2507 with the same API?

Yes. Use the same OpenAI-compatible TrustedRouter base URL and API key, then change only the model id between openai/gpt-oss-120b and qwen/qwen3-235b-a22b-thinking-2507. This makes side-by-side evals possible without maintaining two provider integrations.

More comparisons

All comparisons

Try gpt-oss-120b and Qwen3 235B A22B Thinking 2507 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.