Updated 2 h ago · first seen 11 Sept 2026
model_01M294ZZ8D4B5BVK4Q0K9DY133
- Parameters
- 4.66B
- T2 · 8 h ago
- Context
- 262.1K tokens
- T2 · 6 h ago
- Released
- 27 Feb 2026
- T2 · 7 h ago
- License
- apache-2.0
- T2 · 7 h ago
Specification
- Release date
- 27 Feb 2026
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 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
- Qwen3_5ForConditionalGeneration
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Parameters
- 4.66B
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 8 h agomedium
- Context window
- 262.1K tokens
Source:Artificial AnalysisT2observed 6 h agomedium
- Base model
- Qwen/Qwen3.5-4B-Base
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- File size
- 9.3 GB
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Hugging Face repo
- Qwen/Qwen3.5-4B
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 8 h agomedium
- Pipeline tag
- image-text-to-text
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
- 7,196,884
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Likes
- 914
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
- featherless-ai
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Last modified
- 2026-03-02T00:52:52+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/Qwen3.5-4B/blob/main/LICENSE
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Aa median output tokens per second
- 26.3
Source:Artificial AnalysisT2observed 6 h agomedium
- Downloads all time
- 43,887,848
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Model type
- qwen3_5
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Tags
- transformers, safetensors, qwen3_5, image-text-to-text
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 7 h agomedium
- Weights dtype
- BF16, F32
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
24
Source tiers
T224
Freshest observation
2 h ago
Conflicts
None
Modalities
Modalities unavailable.
Capabilities
Tool calling
Unavailable
Structured output
Unavailable
Reasoning
Yes
Artificial Analysis · T2
Vision
Unavailable
Audio
Unavailable
Fine-tuning available
Unavailable
- Context window
- 262.1K tokens
Source:Artificial AnalysisT2observed 6 h agomedium
| Benchmark | Score | Metric | Config | Evaluated | Source |
|---|---|---|---|---|---|
| τ²-bench | 92.11% | pass^1 | aa_slug=qwen3-5-4b · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Artificial Analysis Intelligence Index | 13.12 | index | aa_slug=qwen3-5-4b · version=4.3 · estimated=true | — | artificialanalysis.aiT2 |
| GPQA | 77.07% | accuracy | aa_slug=qwen3-5-4b · variant=GPQA Diamond · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Humanity's Last Exam | 9.92% | accuracy | aa_slug=qwen3-5-4b · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| IFBench | 51.97% | accuracy | aa_slug=qwen3-5-4b · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| MMMU-Pro | 65.38% | accuracy | aa_slug=qwen3-5-4b · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Terminal-Bench | 25.84% | accuracy | aa_slug=qwen3-5-4b · variant=v2.1 · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Terminal-Bench | 18.18% | accuracy | aa_slug=qwen3-5-4b · 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
No current prices recorded
Price history
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 | 3.2 GB | Yes |
| Apple M3 Ultra | 4bit | — | 3.2 GB | Yes |
| Mac Studio (Apple M5 Ultra) | 4bit | — | 3.2 GB | Yes |
| AMD Instinct MI325X | 4bit | 256 GB | 3.2 GB | Yes |
| AMD Instinct MI300X | 4bit | 192 GB | 3.2 GB | Yes |
| Apple M2 Ultra | 4bit | — | 3.2 GB | Yes |
| NVIDIA B200 | 4bit | 180 GB | 3.2 GB | Yes |
| NVIDIA H200 | 4bit | 141 GB | 3.2 GB | Yes |
| NVIDIA H200 NVL | 4bit | 141 GB | 3.2 GB | Yes |
| Apple M1 Ultra | 4bit | — | 3.2 GB | Yes |
| Apple M3 Max | 4bit | — | 3.2 GB | Yes |
| Apple M4 Max | 4bit | — | 3.2 GB | Yes |
| Mac Studio (Apple M5 Max) | 4bit | — | 3.2 GB | Yes |
| MacBook Pro (Apple M5 Max) | 4bit | — | 3.2 GB | Yes |
| NVIDIA DGX Spark | 4bit | 128 GB | 3.2 GB | Yes |
| Apple M2 Max | 4bit | — | 3.2 GB | Yes |
| NVIDIA H100 NVL | 4bit | 94 GB | 3.2 GB | Yes |
| NVIDIA A100 80GB | 4bit | 80 GB | 3.2 GB | Yes |
| NVIDIA H100 SXM | 4bit | 80 GB | 3.2 GB | Yes |
| Apple M1 Max | 4bit | — | 3.2 GB | Yes |
| Apple M4 Pro | 4bit | — | 3.2 GB | Yes |
| Mac mini (Apple M5 Pro) | 4bit | — | 3.2 GB | Yes |
| MacBook Pro (Apple M5 Pro) | 4bit | — | 3.2 GB | Yes |
| Apple M3 Pro | 4bit | — | 3.2 GB | Yes |
| Apple M1 Pro | 4bit | — | 3.2 GB | Yes |
| Apple M2 Pro | 4bit | — | 3.2 GB | Yes |
| Apple M4 | 4bit | — | 3.2 GB | Yes |
| iMac (Apple M4) | 4bit | — | 3.2 GB | Yes |
| Mac mini (Apple M6) | 4bit | — | 3.2 GB | Yes |
| MacBook Air (Apple M5) | 4bit | — | 3.2 GB | Yes |
| MacBook Pro (Apple M5) | 4bit | — | 3.2 GB | Yes |
| NVIDIA GeForce RTX 5090 | 4bit | 32 GB | 3.2 GB | Yes |
| Apple M2 | 4bit | — | 3.2 GB | Yes |
| Apple M3 | 4bit | — | 3.2 GB | Yes |
| NVIDIA GeForce RTX 3090 | 4bit | 24 GB | 3.2 GB | Yes |
| NVIDIA GeForce RTX 4090 | 4bit | 24 GB | 3.2 GB | Yes |
| Apple M1 | 4bit | — | 3.2 GB | Yes |
Ancestors 1
- Qwen3.5-4B-BaseQwen
This model
Qwen3.5-4B
4.66B params
Quantizations 3
Descendants 0
None recorded.
Papers 0
No papers linked yet.
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 · Aa median output tokens per second
Aa median output tokens per secondmetric.aa_median_output_tokens_per_second1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| 26.3 | → current | current | Artificial AnalysisT2 | 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 →
Qwen3.5-4B scores 9.92% on Humanity's Last Exam
artificial_analysisQwen3.5-4B scores 13.12 on Artificial Analysis Intelligence Index
artificial_analysis
| Source | Document | Type | Tier | Last observed | Snapshots |
|---|---|---|---|---|---|
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/Qwen/Qwen3.5-4B/raw/main/README.md | model_card | T2· Quality secondary | 59 min ago | 1 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/unsloth/Qwen3.5-4B-GGUF | model_page | T2· Quality secondary | 1 h ago | 3 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/Qwen/Qwen3.5-4B | model_page | T2· Quality secondary | 2 h ago | 4 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/Intel/Qwen3.5-4B-int4-AutoRound | model_page | T2· Quality secondary | 2 h ago | 5 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/bartowski/Qwen_Qwen3.5-4B-GGUF | model_page | T2· Quality secondary | 2 h ago | 5 |
| 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 | 6 h ago | 1 |
Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.