Qwen3 Coder 480B: token counter & pricing
Alibaba · approximate, within ±3% of reference · pricing as of 2026-05-31.
- Provider
- Alibaba
- API model ID
Qwen/Qwen3-Coder-480B-A35B-Instruct- Context window
- 131,072 tokens
- Input price
- $2.00 per 1M tokens
- Output price
- $2.00 per 1M tokens
- Tokenizer accuracy
- approximate, within ±3% of reference
- Pricing as of
- 2026-05-31
Open the counter to count tokens for Qwen3 Coder 480B in real time.
What is Qwen3 Coder 480B?
Qwen3 Coder 480B is Alibaba's current flagship code-specialized open-weights model, a 480-billion-parameter Mixture-of-Experts model with 35B active parameters, designed specifically for code generation and editing tasks.
$2 input / $2 output per 1M tokens (Together.ai indicative). Single-rate pricing typical of open-weights inference providers.
How tokens are counted here
Qwen uses a custom BPE tokenizer efficient across English and CJK (Chinese, Japanese, Korean). We approximate in your browser, accurate within roughly ±3% for English. CJK accuracy is similar but less thoroughly validated. Marked ≈±3%.
Pricing notes
$2 input / $2 output per 1M (Together.ai). Single-rate, output and input cost the same per token.
For 1,000 input + 200 output: $0.0024 per call, $2,400 per 1M calls.
131,072-token context window.
When to use Qwen3 Coder 480B
- Current flagship open-weights coding model, measurably outperforms Llama 3.3 and Qwen 2.5 Coder 32B on most coding benchmarks (HumanEval, SWE-Bench, MBPP).
- Coding agents and IDE integrations where you want open-weights with strong code generation.
- Multilingual code, Qwen handles CJK comments and identifiers better than Llama.
When not to use it:
- Pure cost, Qwen 2.5 Coder 32B at $0.80/$0.80 is 2.5× cheaper if your tasks don't need 480B's capacity.
- Closed-model alternatives, Claude Sonnet 4.6 ($3/$15) often wins on reading large existing codebases; GPT-5.3 Codex ($1.75/$14) wins on greenfield generation.
- Self-hosting, 480B requires serious GPU infrastructure; use 32B if hosting yourself.
Common questions
Qwen3 Coder 480B vs Claude Sonnet for code?
Sonnet wins on understanding large existing codebases and producing edits that match local conventions. Qwen3 Coder wins on raw greenfield generation benchmarks. Cost: Sonnet $3/$15, Qwen3 Coder $2/$2, Qwen is dramatically cheaper for output-heavy generation. Choose by workload shape.
Same tokenizer as Qwen 2.5 Coder?
Functionally similar (Qwen family tokenizer). Token counts should be within a few percent across versions. The calculator's ≈±3% confidence label applies to both.
Can I self-host?
Yes, but 480B requires substantial hardware, typical inference setup is multi-GPU (8× H100 or similar). For self-hosting, Qwen 2.5 Coder 32B is a more realistic choice; runs on a single 24GB GPU when quantized.
Compare Qwen3 Coder 480B to other models
- Qwen 2.5 72B (Alibaba, $0.90/$0.90)
- Qwen 2.5 Coder 32B (Alibaba, $0.80/$0.80)
- GPT-4.1 (OpenAI, $2.00/$8.00)
- o3 (OpenAI, $2.00/$8.00)
- Gemini 3.1 Pro (Google, $2.00/$12.00)