OSAIM
Open Source AI Models

QwQ 32B Preview

Qwen's reasoning-focused 'thinking' model. Generates long chains-of-thought before answering, similar to OpenAI's o1 and DeepSeek R1 lineage. Optimised for math and competition-style problem solving.

The Preview tag means Qwen is iterating quickly; later versions may obsolete this one. Useful today for math-heavy workloads where a slow, careful answer is preferred to a fast wrong one.

Parameters
32B
Context length
33K
Modality
text
Released
2024-11-28

Memory & hardware

VRAM (fp16)
64 GB
VRAM (Q4)
19.2 GB
Recommended
H100 80GB or RTX 4090 (Q4)
Quantizations
fp16, q8_0, q4_k_m

License: Apache 2.0

SPDX
Apache-2.0
Commercial use
Yes
Modification
Yes
Redistribution
Yes

Benchmarks

MATH
90.6
MMLU
75.0
GPQA
65.2
Benchmarks last verified 2026-05-18.

Hosted inference pricing

USD per million tokens.

ProviderInputOutput
togetherCheapest$1.20$1.20
Pricing last verified 2026-05-18. Providers update rates frequently; confirm before integrating.

Run it yourself

Drop-in commands for the three most common open-source inference paths. The Ollama tag is a best-effort match against the registry; verify the size variant before pulling.

Run QwQ 32B Preview locally
No official Ollama registry tag for this model — use transformers or vLLM below.
vLLM (production)
vllm serve Qwen/QwQ-32B-Preview
High-throughput hosted inference; one command to expose an OpenAI-compatible HTTP server.
Transformers (Python)
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Qwen/QwQ-32B-Preview")
model = AutoModelForCausalLM.from_pretrained(
    "Qwen/QwQ-32B-Preview", device_map="auto", torch_dtype="auto"
)
Direct PyTorch usage. Pin a torch / cuda version that matches your GPU.
Hugging Face ID: Qwen/QwQ-32B-Preview

Related models

Same family or similar size — useful when shopping around.

Qwen2.5 32B Instruct
32B

32B sweet-spot model: strong reasoning, fits on one H100 in fp16, on a 4090 at Q4. The 32B size in particular hits a quality/cost knee — quality scales with parameters faster than cost up to ~32B, and slower afterwards. Favoured for production chat where 7B isn't sharp enough and where 70B+ would over-spec the hardware budget. Apache 2.0 licence.

Context
128K
License
apache-2-0
VRAM Q4
19.2 GB
Qwen2.5 Coder 32B
32B

Coding-specialised Qwen2.5 32B fine-tune. GPT-4o-class on HumanEval and BigCodeBench at the time of release. Trained on additional code-heavy data with extended pre-training. Apache 2.0. Natural pick for self-hosted coding assistants, code-review automation, and any agent loop that primarily writes code.

Context
128K
License
apache-2-0
VRAM Q4
19.2 GB
Yi 1.5 34B Chat
34B

Bilingual EN/中 34B chat model. Apache 2.0 licensed with strong Chinese-language performance and competitive English chat quality. Good default for bilingual production workloads.

Context
33K
License
apache-2-0
VRAM Q4
20.4 GB
Yi VL 34B
34B

Vision-language variant of Yi 34B. Image-text reasoning via an MLP adapter on a CLIP encoder. Useful for bilingual EN/中 multimodal workloads where the major Western vision-language models underperform on Chinese text in images.

Context
4K
License
apache-2-0
VRAM Q4
20.4 GB
Command R
35B

Cohere's 35B model tuned for RAG and tool use. The open weights are released under CC-BY-NC (commercial use requires the Cohere API). Strong multilingual coverage and a fine-grained RAG-mode output format that makes downstream citation easier.

Context
128K
License
mrl
VRAM Q4
21 GB
Mixtral 8×7B Instruct
46.7B

The mixture-of-experts release that introduced 8 experts of 7B each, 2 active per token. ~13B active parameters with 47B total, which makes per-token inference roughly as fast as a 13B dense model while approaching 70B dense quality. Apache 2.0 weights mean it's still a popular self-hosting choice. Memory footprint is the main constraint — the full 47B parameters must be loaded even though only a quarter are active per token.

Context
33K
License
apache-2-0
VRAM Q4
28 GB