Qwen3 32B
Qwenhuggingface.co/Qwen/Qwen3-32B
Qwen3-32B is a dense 32.8B parameter causal language model from the Qwen3 series, optimized for both complex reasoning and efficient dialogue. It supports seamless switching between a "thinking" mode for...
Updated 1 h ago · first seen 11 Sept 2026
model_01M294WWGS9G50W2QP5DP4KF6Z
- Parameters
- 32.8B
- T2 · 8 h ago
- Context
- 131.1K tokens
- T2 · 5 h ago
- Released
- 27 Apr 2025
- T2 · 3 h ago
- License
- apache-2.0
- T2 · 7 h ago
- Knowledge cutoff
- Mar 2025
- T2 · 8 h ago
Specification
- Release date
- 27 Apr 2025
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 3 h agomedium
- Status
- deprecated
Source:Artificial AnalysisT2observed 7 h agomedium
- Openness
- open-weights
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 8 h agomedium
- License
- apache-2.0
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Architecture
- Qwen3ForCausalLM
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Parameters
- 32.8B
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 8 h agomedium
- Context window
- 131.1K tokens
Source:OpenRouter public model & pricing listingT2observed 5 h agomedium
- Max output
- 16.4K tokens
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- Knowledge cutoff
- Mar 2025
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- Modalities
- text
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- Input modalities
- text
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- Output modalities
- text
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- Tokenizer
- Qwen3
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- File size
- 65.5 GB
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Hugging Face repo
- Qwen/Qwen3-32B
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- Pipeline tag
- text-generation
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 8 h agomedium
- Model card
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 8 h agomedium
- Downloads
- 5,273,218
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Likes
- 745
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Gated
- No
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Hf inference providers
- deepinfra, featherless-ai, groq, novita, nscale
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Instruct type
- qwen3
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- Last modified
- 2025-07-26T03:45:22+00:00
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 8 h agomedium
- Library name
- transformers
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- License url
- https://huggingface.co/Qwen/Qwen3-32B/blob/main/LICENSE
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Aa median output tokens per second
- 104.7
Source:Artificial AnalysisT2observed 7 h agomedium
- Downloads all time
- 56,250,212
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Model type
- qwen3
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Openrouter id
- qwen/qwen3-32b
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- Openrouter listed at
- 28 Apr 2025
Source:OpenRouter public model & pricing listingT2observed 1 h agomedium
- Supported parameters
- frequency_penalty, include_reasoning, logit_bias, max_tokens, min_p, presence_penalty, reasoning, repetition_penalty, response_format, seed, stop, structured_outputs, temperature, tool_choice, tools, top_k, top_p
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- Tags
- transformers, safetensors, qwen3, text-generation
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Weights dtype
- BF16
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →
Provenance
Attributed facts
37
Source tiers
T237
Freshest observation
1 h ago
Conflicts
None
Modalities
- Modalities
- text
- Input
- text
- Output
- text
Capabilities
Tool calling
Yes
OpenRouter public model & pricing listing · T2
Structured output
Yes
OpenRouter public model & pricing listing · T2
Reasoning
Yes
OpenRouter public model & pricing listing · T2
Vision
Unavailable
Audio
Unavailable
Fine-tuning available
Unavailable
- Context window
- 131.1K tokens
Source:OpenRouter public model & pricing listingT2observed 5 h agomedium
- Max output
- 16.4K tokens
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- Knowledge cutoff
- Mar 2025
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
- Tokenizer
- Qwen3
Source:OpenRouter public model & pricing listingT2observed 8 h agomedium
| Benchmark | Score | Metric | Config | Evaluated | Source |
|---|---|---|---|---|---|
| τ²-bench | 29.82% | pass^1 | aa_slug=qwen3-32b-instruct-reasoning · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Aider polyglot | 83.6% | percent_cases_well_formed | date=2025-05-08 · command=aider --model openrouter/qwen/qwen3-32b · dirname=2025-05-08-03-20-24--qwen3-32b-default · versions=0.82.4.dev | 8 May 2025 | aider.chatT2 |
| Aider polyglot | 40% | pass_rate_2 | date=2025-05-08 · command=aider --model openrouter/qwen/qwen3-32b · dirname=2025-05-08-03-20-24--qwen3-32b-default · versions=0.82.4.dev | 8 May 2025 | aider.chatT2 |
| Artificial Analysis Intelligence Index | 7.22 | index | aa_slug=qwen3-32b-instruct-reasoning · version=4.3 · estimated=false | — | artificialanalysis.aiT2 |
| GPQA | 66.77% | accuracy | aa_slug=qwen3-32b-instruct-reasoning · variant=GPQA Diamond · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Humanity's Last Exam | 7.41% | accuracy | aa_slug=qwen3-32b-instruct-reasoning · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| IFBench | 36.33% | accuracy | aa_slug=qwen3-32b-instruct-reasoning · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| SciCode | 36% | accuracy | aa_slug=qwen3-32b-instruct-reasoning · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Terminal-Bench | 0% | accuracy | aa_slug=qwen3-32b-instruct-reasoning · variant=v4.0 · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Terminal-Bench | 5.24% | accuracy | aa_slug=qwen3-32b-instruct-reasoning · variant=v2.1 · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Terminal-Bench | 3.03% | accuracy | aa_slug=qwen3-32b-instruct-reasoning · variant=hard · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
Scores are reported as published, with their evaluation configuration (harness, prompting, judge). Results with different configs are not directly comparable — see methodology.
Current prices
| Provider | Input / 1M | Output / 1M | Cached in | Batch in / out | Context | Observed | Source |
|---|---|---|---|---|---|---|---|
| OpenRouterqwen/qwen3-32b | $0.08 | $0.28 | — | — | 41K | 1 h ago | openrouter.aiT2 |
USD per 1M tokens as published by each provider (USD). Rows are append-only: every change is kept in the history below.
Available through 1
Price history
Input price · USD / 1M tokens
Output price · USD / 1M tokens
- OpenRouter
Memory need = bytes per parameter (4-bit ≈ 0.5 × 1.15 overhead, 8-bit 1.0, fp16 2.0) + a KV-cache allowance. Not a measurement.
| Hardware | Quantization | Memory | Est. need | Fits |
|---|---|---|---|---|
| NVIDIA DGX B200 | 4bit | 1,440 GB | 19.3 GB | Yes |
| Apple M3 Ultra | 4bit | — | 19.3 GB | Yes |
| Mac Studio (Apple M5 Ultra) | 4bit | — | 19.3 GB | Yes |
| AMD Instinct MI325X | 4bit | 256 GB | 19.3 GB | Yes |
| AMD Instinct MI300X | 4bit | 192 GB | 19.3 GB | Yes |
| Apple M2 Ultra | 4bit | — | 19.3 GB | Yes |
| NVIDIA B200 | 4bit | 180 GB | 19.3 GB | Yes |
| NVIDIA H200 | 4bit | 141 GB | 19.3 GB | Yes |
| NVIDIA H200 NVL | 4bit | 141 GB | 19.3 GB | Yes |
| Apple M1 Ultra | 4bit | — | 19.3 GB | Yes |
| Apple M3 Max | 4bit | — | 19.3 GB | Yes |
| Apple M4 Max | 4bit | — | 19.3 GB | Yes |
| Mac Studio (Apple M5 Max) | 4bit | — | 19.3 GB | Yes |
| MacBook Pro (Apple M5 Max) | 4bit | — | 19.3 GB | Yes |
| NVIDIA DGX Spark | 4bit | 128 GB | 19.3 GB | Yes |
| Apple M2 Max | 4bit | — | 19.3 GB | Yes |
| NVIDIA H100 NVL | 4bit | 94 GB | 19.3 GB | Yes |
| NVIDIA A100 80GB | 4bit | 80 GB | 19.3 GB | Yes |
| NVIDIA H100 SXM | 4bit | 80 GB | 19.3 GB | Yes |
| Apple M1 Max | 4bit | — | 19.3 GB | Yes |
| Apple M4 Pro | 4bit | — | 19.3 GB | Yes |
| Mac mini (Apple M5 Pro) | 4bit | — | 19.3 GB | Yes |
| MacBook Pro (Apple M5 Pro) | 4bit | — | 19.3 GB | Yes |
| Apple M3 Pro | 4bit | — | 19.3 GB | Yes |
| Apple M1 Pro | 4bit | — | 19.3 GB | Yes |
| Apple M2 Pro | 4bit | — | 19.3 GB | Yes |
| Apple M4 | 4bit | — | 19.3 GB | Yes |
| iMac (Apple M4) | 4bit | — | 19.3 GB | Yes |
| Mac mini (Apple M6) | 4bit | — | 19.3 GB | Yes |
| MacBook Air (Apple M5) | 4bit | — | 19.3 GB | Yes |
| MacBook Pro (Apple M5) | 4bit | — | 19.3 GB | Yes |
| NVIDIA GeForce RTX 5090 | 4bit | 32 GB | 19.3 GB | Yes |
| Apple M2 | 4bit | — | 19.3 GB | Yes |
| Apple M3 | 4bit | — | 19.3 GB | Yes |
| NVIDIA GeForce RTX 3090 | 4bit | 24 GB | 19.3 GB | Yes |
| NVIDIA GeForce RTX 4090 | 4bit | 24 GB | 19.3 GB | Yes |
| Apple M1 | 4bit | — | 19.3 GB | No |
Ancestors 0
None recorded.
This model
Qwen3 32B
32.8B params
Quantizations 0
None recorded.
Descendants 1
Papers 2
- arXiv:2309.00071Active35
- arXiv:2505.09388Active35
Repositories 0
No repositories linked yet.
As of
Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.
Claim history · Weights dtype
Weights dtypeweights_dtype1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| BF16 | → current | current | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →
Qwen3 32B: context length changed from 32768 to 131072
Context window32.8K tokens→131.1K tokensopenrouterQwen3 32B scores 7.22 on Artificial Analysis Intelligence Index
artificial_analysisQwen3 32B: context length changed from 131072 to 32768
Context window131.1K tokens→32.8K tokensartificial_analysisQwen3 32B: release date changed from 2025-04-27 to 2025-04-28
Release date27 Apr 2025→28 Apr 2025openrouterQwen3 32B: release date changed from 2025-04-28 to 2025-04-27
Release date28 Apr 2025→27 Apr 2025huggingfaceOpenRouter lists Qwen3 32B at $0.08 in / $0.28 out per 1M tokens
openrouter
| Source | Document | Type | Tier | Last observed | Snapshots |
|---|---|---|---|---|---|
| Qwen — official blog | qwenlm.github.io/blog/qwen3 | news | T1· Official | 8 h ago | 1 |
| OpenRouter public model & pricing listing | openrouter.ai/api/v1/models | catalogue | T2· Quality secondary | 1 h ago | 5 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/Qwen/Qwen3-32B/raw/main/README.md | model_card | T2· Quality secondary | 1 h ago | 1 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/Qwen/Qwen3-32B | model_page | T2· Quality secondary | 2 h ago | 4 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/amazon/GKA-primed-HQwen3-32B-Reasoner | model_page | T2· Quality secondary | 2 h ago | 6 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/models?author=Qwen&p=0&sort=downloads | listing | T2· Quality secondary | 2 h ago | 6 |
| Artificial Analysis | artificialanalysis.ai/leaderboards/models | leaderboard | T2· Quality secondary | 7 h ago | 1 |
| Aider polyglot leaderboard (data file on GitHub) | raw.githubusercontent.com/Aider-AI/aider/main/aider/website/_data/polyglot_leaderboard.yml | leaderboard | T2· Quality secondary | 7 h ago | 1 |
Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.