50 models7 providersprices scraped nightly — no estimates
alibaba · qwen familyVerified May 16, 2026

Qwen 2.5 Coder 32B Instruct

32B params · 131K context · qwen · released 2024-11 · 1 provider

Qwen 2.5 Coder 32B Instruct is a code-specialized fine-tune of the Qwen 2.5 32B base, released by Alibaba in November 2024 with support for 92 programming languages across a 131K context window. On HumanEval, LiveCodeBench, and MultiPL-E it sits alongside DeepSeek Coder V2 and Codestral as a credible alternative for code completion and agentic coding pipelines, at 32B parameters rather than larger model classes. Hosted primarily on DeepInfra and a handful of other providers, it offers a cost-effective backbone for CI-integrated code generation or IDE completions where latency tolerances are looser than sub-second. The Qwen license permits commercial deployment. Broader provider adoption is likely as code-model demand grows; watch for pricing to tighten against Codestral as supply catches up.

Cheapest right now
DeepInfra$0.25/1M out
sticky while scrolling · verified May 16, 2026
Cheapest now
$0.25
DeepInfra · fp16 · $/1M out
90-day price Δ
cheapest-host trend
Fastest TTFT
300 ms
DeepInfra · 280 tok/s
Providers
1
indexed providers
✓ verified May 16, 2026
#Provider$/1M in$/1M outTTFT p50tok/sQuant
1DeepInfraCHEAPEST$0.12$0.25300 ms280fp16go ↗

Prices verified May 16, 2026 by scraping 1 provider. Methodology · Raw data