- Size
- 8.2B parameterscounted from the safetensors header on the Hub
- Licence
- custom, read it, Apache License 2.0Our reading, check the licence: A custom licence we have not read for this model: other makers use the same name for different texts. Read the model's own licence file before you use it.Licence text
- On the Hub since
- 27 Apr 2025the date the repo was created
- Context length
- 40k tokensfrom the model's config
- Fine-tune
- of Qwen/Qwen3-8B-BaseQwen3ForCausalLM
- On Hugging Face
- Qwen/Qwen3-8B6.1M downloads in the last 30 days, 2.2k likes
Scores
Epoch's name: qwen3-8b. Best closed model on this set: gpt-6.2-astra_max, 96.8% (OpenAI). All 341 results
Epoch's name: qwen3-8b. Best closed model on this set: claude-fable-5_high, 100.0% (Anthropic). All 318 results
- default setting0.0%via Epoch AI
Epoch's name: qwen3-8b. Best closed model on this set: gemini-4-carbon_high, 36.0% (Google DeepMind). All 271 results
- default setting59.7%conceptualreasoning.ai
Epoch's name: qwen3-8b. Best closed model on this set: claude-fable-5-2_unknown, 99.2% (Anthropic). All 241 results
- default setting62.1%Fiction.live leaderboard
Epoch's name: chutes/Qwen3-8B. Best closed model on this set: gpt-5-2025-08-07_medium, 97.2% (OpenAI). All 61 results
- default setting8.8%3.7 to 13.9conceptualreasoning.ai
Epoch's name: qwen3-8b. Best closed model on this set: claude-fable-5-2_xhigh, 70.6% (Anthropic). All 203 results
Claimed by the maker Claimed
The model card has no claimed scores in a machine-readable form (the card's model-index), so none are shown. The maker may give scores in the card text: read the card.
Other views, which we link to and do not copy: look this model up on Artificial Analysis and the Arena leaderboard.
Independent scores from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Each row links its own source. What the badges mean.
Run it yourself
Memory needed = weights + context cache + 0.50 GB for the runtime. Context cache at 8k tokens: 1.1 GB, worked out from the model's config (Qwen/Qwen3-8B), at 16-bit precision. Choose a machine above to see a verdict for each format. Method
GGUF for llama.cpp, Ollama, LM Studio and others
Qwen/Qwen3-8B-GGUF (the maker's own conversion), 852k downloads. Sizes are the real files on the Hub, all parts added up.
| Format | Size | Memory needed | Get it |
|---|---|---|---|
| Q4_K_M | 4.7 GBfile size | 6.3 GB4.7 + 1.1 + 0.50 | Commandsllama.cpp llama-server -hf Qwen/Qwen3-8B-GGUF:Q4_K_MOllama ollama run hf.co/Qwen/Qwen3-8B-GGUF:Q4_K_MDownload hf download Qwen/Qwen3-8B-GGUF --include "Qwen3-8B-Q4_K_M.gguf" --local-dir Qwen3-8B-GGUF |
| Q5_0 | 5.3 GBfile size | 7.0 GB5.3 + 1.1 + 0.50 | Commandsllama.cpp llama-server -hf Qwen/Qwen3-8B-GGUF:Q5_0Ollama ollama run hf.co/Qwen/Qwen3-8B-GGUF:Q5_0Download hf download Qwen/Qwen3-8B-GGUF --include "Qwen3-8B-Q5_0.gguf" --local-dir Qwen3-8B-GGUF |
| Q5_K_M | 5.4 GBfile size | 7.1 GB5.4 + 1.1 + 0.50 | Commandsllama.cpp llama-server -hf Qwen/Qwen3-8B-GGUF:Q5_K_MOllama ollama run hf.co/Qwen/Qwen3-8B-GGUF:Q5_K_MDownload hf download Qwen/Qwen3-8B-GGUF --include "Qwen3-8B-Q5_K_M.gguf" --local-dir Qwen3-8B-GGUF |
| Q6_K | 6.3 GBfile size | 7.9 GB6.3 + 1.1 + 0.50 | Commandsllama.cpp llama-server -hf Qwen/Qwen3-8B-GGUF:Q6_KOllama ollama run hf.co/Qwen/Qwen3-8B-GGUF:Q6_KDownload hf download Qwen/Qwen3-8B-GGUF --include "Qwen3-8B-Q6_K.gguf" --local-dir Qwen3-8B-GGUF |
| Q8_0 | 8.1 GBfile size | 9.7 GB8.1 + 1.1 + 0.50 | Commandsllama.cpp llama-server -hf Qwen/Qwen3-8B-GGUF:Q8_0Ollama ollama run hf.co/Qwen/Qwen3-8B-GGUF:Q8_0Download hf download Qwen/Qwen3-8B-GGUF --include "Qwen3-8B-Q8_0.gguf" --local-dir Qwen3-8B-GGUF |
MLX for Apple silicon
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| MLX 4bitmlx-community/Qwen3-8B-4bit | 4.3 GBfile size | 5.9 GB4.3 + 1.1 + 0.50 | Download page |
| MLX 8bitlmstudio-community/Qwen3-8B-MLX-8bit | 8.1 GBfile size | 9.7 GB8.1 + 1.1 + 0.50 | Download page |
Original weights size from the safetensors header, for vLLM, SGLang, Transformers
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| BF16original weights | 15.3 GBfile size | 16.9 GB15.3 + 1.1 + 0.50 |
hf download Qwen/Qwen3-8B --local-dir Qwen3-8BOther conversions (18; sizes not read)
- Qwen/Qwen3-8B-AWQ AWQ, by the maker, 949k downloads
- Qwen/Qwen3-8B-FP8 FP8, by the maker, 297k downloads
- unsloth/Qwen3-8B-unsloth-bnb-4bit bitsandbytes, 4bit, 210k downloads
- nvidia/Qwen3-8B-NVFP4 NVFP4, 118k downloads
- mobilint/Qwen3-8B Other, 27k downloads
- unsloth/Qwen3-8B-bnb-4bit bitsandbytes, 4bit, 22k downloads
- RedHatAI/Qwen3-8B-FP8-dynamic FP8, 5.5k downloads
- RedHatAI/Qwen3-8B-quantized.w4a16 Other, w4a16, 5.2k downloads
- nvidia/Qwen3-8B-FP8 FP8, 3.8k downloads
- OpenVINO/Qwen3-8B-int4-ov Other, int4, 2.4k downloads
- OpenVINO/Qwen3-8B-int4-cw-ov Other, int4, 2.0k downloads
- JunHowie/Qwen3-8B-GPTQ-Int4 GPTQ, int4, 969 downloads
- RedHatAI/Qwen3-8B-NVFP4 NVFP4, 851 downloads
- Minguinho-zeze/Qwen3-8B-NVFP4A16-RTN NVFP4, 785 downloads
- Minguinho-zeze/Qwen3-8B-NVFP4-RTN-WikiText2 NVFP4, 746 downloads
- JunHowie/Qwen3-8B-GPTQ-Int8 GPTQ, int8, 669 downloads
- pytorch/Qwen3-8B-AWQ-INT4 AWQ, int4, 582 downloads
- chris320211/autoround-qwen3-8b-vastrtx3090-vllm GPTQ, 342 downloads
The Hub tags 100 or more quantised repos as conversions of this model; we read the most downloaded and show the maker's own first.
Run it by API
No provider on Hugging Face Inference Providers serves this model (as of 19 Nov 2026, 13:58 UTC). Other hosts may; we only list what the router reports.
Not built yet: one key for every open model, with BlockRun, and our own tested-quality score per provider. Until then, precision and measured speed are the best guide to a provider here.
Fine-tunes of this model
Repos that declare this model as their base, most downloaded first. Listing is not a recommendation; unreviewed ones have not been checked.
- Qwen/Qwen3Guard-Gen-8B8.2B100kUnreviewed
- unsloth/Qwen3-8B8.2B61kUnreviewed
- allenai/Molmo2-8B8.7B51kUnreviewed
- huihui-ai/Huihui-Qwen3-8B-abliterated-v28.2B15kUnreviewed
- DreamFast/qwen3-8b-heretic8.2B10kUnreviewed
- Open-Bee/Bee-8B-RL8.7B8.1kUnreviewed
- allenai/MolmoPoint-8B8.7B5.5kUnreviewed
- nvidia/Nemotron-Orchestrator-8B8.2B5.2kUnreviewed
- Qwen/WebWorld-32B3.5kUnreviewed
- openbmb/MathForm-8B8.2B2.4kUnreviewed
- allenai/MolmoWeb-4B4.9B2.3kUnreviewed
- nvidia/Qwen3-Nemotron-8B-BRRM2.1kUnreviewed
- huihui-ai/Qwen3-8B-abliterated8.2B2.0kUnreviewed
- EleutherAI/qwen3-8b-djinnsdf-dolci8.2B1.9kUnreviewed
- LastMachine/Qwen3-8B-ZeroZero-Heretic8.2B1.7kUnreviewed
- allenai/AstaBrief_8B_SFT1.4kUnreviewed
- Qwen/WebWorld-8B8.2B1.3kUnreviewed
- Qwen/WebWorld-14B14.8B1.1kUnreviewed
- tencent/ContextPilot-8B8.2B879Unreviewed
- allenai/MolmoWeb-8B8.7B857Unreviewed
Planned for this page Not built yet
- Try it in private chat
- A link to Solana AI's confidential chat for the models it serves. This model is not listed as served there.
- Fine-tune it
- Fine-tuning a few base models from this page and publishing the result here.
- Creators and earnings
- Pages for people who publish fine-tunes and adapters of this model, and what they earn.
- Fund and bounties
- Bounties for new open models, paid from a public fund. How the fund will work.
- Contribute compute
- How to give compute to open model work, with Akash.
Model data from the Hugging Face Hub 19 Nov 2026, 13:43 UTC. The repo. Found a mistake on this page? Tell us.