// head_to_head
GPT-5.2-Codex vs Qwen3.6 Plus
Published LiveBench scores across all seven categories, live list pricing, context windows, and the measured cost of a point of capability — for both models, side by side.
GPT-5.2-Codex scores higher, Qwen3.6 Plus costs less — it depends on your workload.
GPT-5.2-Codex is ahead by 5.1 points overall, and Qwen3.6 Plus lists 6.6× cheaper per blended million tokens. Whether 5.1 points is worth that depends on how much a wrong answer costs you. The two cost measures disagree here, which is worth knowing: Qwen3.6 Plus has the lower sticker price, but GPT-5.2-Codex earns each point of capability for less — $0.1001 against $0.1262 — because per-token rates do not predict how many tokens a model actually spends on a task.
openai
GPT-5.2-Codex
- Blended / 1M
- $4.81
- Context
- 400K
- Released
- Jan 14, 2026
- Overall score
- 74.0
qwen
Qwen3.6 Plus
- Blended / 1M
- $0.731
- Context
- 1M
- Released
- Apr 2, 2026
- Overall score
- 68.9
Specs and pricing
| Metric | GPT-5.2-Codex | Qwen3.6 Plus |
|---|---|---|
| LiveBench overall Mean of the seven LiveBench category scores, 0–100. Higher is better. | 74.0win | 68.9 |
| Cost per point Measured benchmark spend divided by overall score — dollars per point of capability. | $0.1001win | $0.1262 |
| Blended price / 1M 3:1 input:output mix, the usual shape of production traffic. | $4.81 | $0.731win |
| Input price / 1M | $1.75 | $0.325win |
| Output price / 1M | $14.00 | $1.95win |
| Cached input / 1M Price of an input token served from the prompt cache, where the provider publishes one. | $0.175 | — |
| Context window | 400K | 1Mwin |
| Max output tokens | 128Kwin | 66K |
Benchmarks by category
An overall score averages away the thing you probably care about. These are the seven LiveBench categories scored separately — GPT-5.2-Codex on top, Qwen3.6 Plus below, both out of 100.
What each one costs to run
Per-token prices are hard to feel. These are monthly list costs for both models across five workload shapes, using each provider's published cached-input rate where there is one.
| Workload | GPT-5.2-Codex | Qwen3.6 Plus |
|---|---|---|
| Support chatbot 1.2K in / 400 out × 200K requests | $1426.60/mo | $234.00/mo |
| RAG assistant 8K in / 600 out × 100K requests | $1610.00/mo | $377.00/mo |
| Coding agent 40K in / 4K out × 20K requests | $1638.00/mo | $416.00/mo |
| Document extraction 20K in / 1.5K out × 50K requests | $2721.25/mo | $471.25/mo |
| Bulk classification 500 in / 20 out × 5M requests | $4987.50/mo | $1007.50/mo |
Which should you pick?
You are running this at volume
GPT-5.2-Codex
Lowest measured cost per point of capability at $0.1001 per point — the gap compounds with every request.
Quality matters more than the bill
GPT-5.2-Codex
Highest overall LiveBench score of the two at 74.0.
The workload is coding or agentic work
GPT-5.2-Codex
Leads on agentic coding — 49.4 against 41.4.
You need to fit large documents in one call
Qwen3.6 Plus
Wider context window — 1M against 400K.
GPT-5.2-Codex vs Qwen3.6 Plus FAQ
Which is better, GPT-5.2-Codex or Qwen3.6 Plus?
GPT-5.2-Codex scores higher, Qwen3.6 Plus costs less — it depends on your workload. GPT-5.2-Codex is ahead by 5.1 points overall, and Qwen3.6 Plus lists 6.6× cheaper per blended million tokens. Whether 5.1 points is worth that depends on how much a wrong answer costs you. The two cost measures disagree here, which is worth knowing: Qwen3.6 Plus has the lower sticker price, but GPT-5.2-Codex earns each point of capability for less — $0.1001 against $0.1262 — because per-token rates do not predict how many tokens a model actually spends on a task.
Is GPT-5.2-Codex cheaper than Qwen3.6 Plus?
Qwen3.6 Plus is cheaper. On a 3:1 input:output blend, GPT-5.2-Codex lists at $4.81 per million tokens and Qwen3.6 Plus at $0.731 — Qwen3.6 Plus is 6.6× cheaper. Input and output are priced separately — GPT-5.2-Codex charges $1.75 in and $14.00 out, Qwen3.6 Plus charges $0.325 and $1.95 — so the model that looks cheaper flips depending on how output-heavy your workload is.
GPT-5.2-Codex vs Qwen3.6 Plus: which scores higher on benchmarks?
GPT-5.2-Codex scores 74.0 and Qwen3.6 Plus scores 68.9 overall on LiveBench, the mean of its seven categories. That is a 5.1-point lead for GPT-5.2-Codex. Category scores differ from the overall figure — a model can lead on reasoning and trail on coding, which the per-category table above breaks out.
Which gives better value for money, GPT-5.2-Codex or Qwen3.6 Plus?
GPT-5.2-Codex. Cost per point divides the measured dollars LiveBench spent running the benchmark by the score it earned, so it captures something token pricing misses: a reasoning model can emit many times more tokens than its per-token rate suggests. GPT-5.2-Codex works out at $0.1001 per point and Qwen3.6 Plus at $0.1262.
Does GPT-5.2-Codex or Qwen3.6 Plus have a bigger context window?
Qwen3.6 Plus has the larger context window: 400K for GPT-5.2-Codex against 1M for Qwen3.6 Plus. Note that a window you can fill is not a window you should fill — retrieval quality usually degrades well before the limit, and you pay for every token you put in it.
Do GPT-5.2-Codex and Qwen3.6 Plus support prompt caching?
GPT-5.2-Codex publishes a cached-input rate of $0.175 per million tokens against a full input rate of $1.75. The catalogue lists no separate cached rate for Qwen3.6 Plus, which means the provider does not price it separately here — not that caching is unavailable.
Related comparisons
How these numbers are produced
- Price — provider list price from OpenRouter, refreshed every 15 minutes. “Blended” is a 3:1 input:output mix.
- Scores — LiveBench release 2026-06-25, using their own category map. Each model shows its strongest published run. A blank means “not evaluated”, never “bad”.
- Cost per point — the measured dollars LiveBench spent on the run, divided by the score it earned.
- “Win” — awarded only past a threshold: one full point on a benchmark score, 10% on a price, 25% on a context window. Anything tighter reports as a tie, because effort settings alone move a LiveBench score by more than that.
Published benchmarks rank models on someone else's tasks. Before committing, see LLM & agent evaluation for building an eval on your own.