Head-to-head showdown: Cerebras — Qwen 3 32B ($0.40 in / $0.80 out per 1M) vs GPT-5.4 mini ($0.75 in / $4.50 out per 1M). Cerebras — Qwen 3 32B is 4.4× cheaper across standard token mixes, with 128K vs 256K context windows.
qwen-3-32b-cerebras · cerebras
gpt-5.4-mini · openai
Benchmark
| Workload Scenario | Cerebras — Qwen 3 32B | GPT-5.4 mini | Price Delta |
|---|---|---|---|
| 1M input tokens (raw text) | $0.40 | $0.75 | −$0.35 |
| 1M output tokens (generation) | $0.80 | $4.50 | −$3.70 |
| 1M tokens · 70% input / 30% output mix | $0.52 | $1.875 | −$1.355 |
| Standard chat turn (4K in / 800 out, 50% cached) | $0.00224 | $0.00525 | −$0.00301 |
| Monthly scale (10K requests / day) | $672.00 | $1,575.00 | −$903.00 |
Negative difference = Cerebras — Qwen 3 32B is cheaper. Positive = GPT-5.4 mini is cheaper.
Scaling Curve
Total cost of a token volume at 70/30 input/output split (uncached). Log-log scale.
Verdict
Cerebras — Qwen 3 32B is cheaper on both input and output rates, so it costs less at every input/output mix. Price alone still isn't the whole decision: capability, latency and context limits (Cerebras — Qwen 3 32B: 128K, GPT-5.4 mini: 256K) may justify the premium for your task.
FAQ
On input, Cerebras — Qwen 3 32B is cheaper ($0.40/M vs $0.75/M). On output, Cerebras — Qwen 3 32B is cheaper ($0.80/M vs $4.50/M). The same model is cheaper on both sides, so it wins at every mix.
A chat-style request (4,000 input + 800 output tokens, 50% cached) costs $0.00224 on Cerebras — Qwen 3 32B and $0.00525 on GPT-5.4 mini — Cerebras — Qwen 3 32B is 2.3× cheaper for that workload.
Cerebras — Qwen 3 32B supports 128,000 tokens (8.2K max output); GPT-5.4 mini supports 256,000 (32.8K max output). GPT-5.4 mini fits 2.0× more context, which matters for long documents and agents.