Up to date
Qwen Apache 2.0 out 2 years ago
14.8B parameters, text in, text out, 32K tokens of context. 1 provider serves it.
Verified 2026-10-25 Rank #127 in LLMs · score 94.8 method v1 Sources 3 Downloads 30d 786.4k · #133 of 619 tracked
Changed this week: 3 new quantized builds on the Hub (204 in all).2026-10-24
Picked by rule from 31 tracked pieces: the lab's own post, the most shared, a field report, the latest.
Qwen2.5 introduces dense decoder-only models pretrained on up to 18 trillion tokens, available in sizes from 0.5B to 72B parameters.
The 2 picks above, plus these 23. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.
Live hosted endpoints, cheapest first, USD per million tokens. Ties break on output price, then provider name.
| # | Cheapest live row | In /1M | Out /1M | Context | Quant | Seen |
|---|---|---|---|---|---|---|
| 1 | Featherless | — | — | — | — | Oct 25 |
Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.
Served by 1 API provider; heavily run locally (204 quantized builds); a big base for builders (958 fine-tunes, adapters and merges).
Top Spaces running it: allenai/reward-bench ♥ 438 · alexnasa/Wan2.2-Animate-ZEROGPU ♥ 365 · multimodalart/self-forcing ♥ 333 · fffiloni/Meigen-MultiTalk ♥ 276
Weights only, an estimate: leave room for the KV cache. Method
About 8.9 GB at INT4, 18 GB at FP8, 35 GB at BF16: weights only, so leave room for the KV cache.
Estimate params × bytes per param × 1.2 · how we estimate
Weights, every dated article, and repo commits on one rail.
Teal is the weights (2024-09-16), amber triangles are 3 dated articles (each links to it), grey ticks are repo commits.
1,162 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.
From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.
Everything else, folded: the full tables, every source.
Qwen2.5-14B-Instruct is an instruction-tuned causal language model with 14.8B parameters.[1][2] The model supports a context length of up to 131,072 tokens and is released under the Apache 2.0 license.[3][4]
Generated 2026-10-07 15:01 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.
Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.
Listed in LLMs
| Pipeline tag | text-generation |
|---|---|
| Library | transformers |
| Architectures | Qwen2ForCausalLM |
| Model type | qwen2 |
| Dtype | bfloat16 |
| Params by dtype | BF16: 14,770,033,664 |
| Size band | L |
| MoE | no |
| Context length | 32,768 (config.json max_position_embeddings) |
| Languages | en |
| Papers | arXiv:2309.00071, arXiv:2407.10671 |
| License | Apache 2.0 (apache-2.0) |
| Hub snapshot | 2026-10-25 |
vLLM: supported · Qwen2ForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · Qwen2ForCausalLM · 773cb938fe9484a8f7e7e46fb1b1fe96412c59ed
vLLM registry (7050ef881ff0835f6890a47db1859a24c65481d6) · SGLang registry (773cb938fe9484a8f7e7e46fb1b1fe96412c59ed) · Matched by architecture class name against each project's registry on main; 'supported' means the architecture is registered, not that this checkpoint was tested.
Weights-only memory for Qwen2.5-14B-Instruct: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 35 GB · Single high-end GPU (<80 GB) |
|---|---|
| FP8 (1 byte/param) | 18 GB · Single GPU (<24 GB) |
| INT4 (0.5 bytes/param) | 8.9 GB · Single GPU (<24 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
Base model: Qwen/Qwen2.5-14B
quantized: 204finetune: 448merge: 79adapter: 431
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| Qwen/Qwen2.5-14B-Instruct-AWQ | quantized | AWQ | 14.8B | L | apache-2.0 | Qwen | 52 | 17.9M |
| Qwen/Qwen2.5-14B-Instruct-GPTQ-Int4 | quantized | GPTQ | 14.8B | L | apache-2.0 | Qwen | 33 | 1.8M |
| unsloth/Qwen2.5-14B-Instruct | finetune | — | 14.8B | L | apache-2.0 | unsloth | 11 | 1.5M |
| unsloth/Qwen2.5-14B-Instruct-unsloth-bnb-4bit | quantized | -bnb | 15.2B | L | apache-2.0 | unsloth | 2 | 824.5k |
| bartowski/Qwen2.5-14B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | bartowski | 74 | 798.4k |
| unsloth/Qwen2.5-14B-Instruct-bnb-4bit | quantized | -bnb | 15.2B | L | apache-2.0 | unsloth | 9 | 617.3k |
| Qwen/Qwen2.5-14B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | Qwen | 77 | 526.7k |
| Qwen/Qwen2.5-14B-Instruct-GPTQ-Int8 | quantized | GPTQ | 14.8B | L | apache-2.0 | Qwen | 31 | 445k |
| lmstudio-community/Qwen2.5-14B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | lmstudio-community | 15 | 190.8k |
| oxyapi/oxy-1-small | finetune | — | 14.8B | L | apache-2.0 | oxyapi | 86 | 144.3k |
| bytedance-research/ChatTS-14B | finetune | — | — | — | apache-2.0 | bytedance-research | 146 | 100.2k |
| rombodawg/Rombos-LLM-V2.6-Qwen-14b | finetune | — | 14.8B | L | apache-2.0 | rombodawg | 54 | 37.3k |
| TheRains/Qwen2.5-14B-Instruct-Q4_K_M-GGUF | quantized | GGUF | — | — | apache-2.0 | TheRains | 4 | 34.1k |
| NCSOFT/VARCO-VISION-14B | finetune | — | 15.2B | L | cc-by-nc-4.0 | NCSOFT | 39 | 30.5k |
| lmstudio-community/Qwen2.5-14B-Instruct-MLX-4bit | quantized | MLX | 14.8B | L | apache-2.0 | lmstudio-community | 3 | 32k |
| v000000/Qwen2.5-14B-Gutenberg-1e-Delta | finetune | — | 14.8B | L | apache-2.0 | v000000 | 4 | 30.9k |
| v000000/Qwen2.5-Lumen-14B | finetune | — | 14.8B | L | apache-2.0 | v000000 | 21 | 29.5k |
| Unbabel/M-Prometheus-14B | finetune | — | 14.8B | L | other | Unbabel | 12 | 29.5k |
| tanliboy/lambda-qwen2.5-14b-dpo-test | finetune | — | 14.8B | L | apache-2.0 | tanliboy | 10 | 28k |
| v000000/Qwen2.5-14B-Gutenberg-Instruct-Slerpeno | merge | — | 14.8B | L | apache-2.0 | v000000 | 7 | 27.8k |
| Lambent/qwen2.5-reinstruct-alternate-lumen-14B | finetune | — | 14.8B | L | — | Lambent | 3 | 27.3k |
| legraphista/Qwen2.5-14B-Instruct-IMat-GGUF | quantized | GGUF | — | — | apache-2.0 | legraphista | 0 | 24.1k |
| mradermacher/Qwen2.5-14B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | mradermacher | 1 | 22.5k |
| eriksalt/qwen2.5-14b-instruct-reddit-rpg-classifier | quantized | GGUF | — | — | apache-2.0 | eriksalt | 0 | 22.2k |
| BytedTsinghua-SIA/RL-MemoryAgent-14B | finetune | — | 14.8B | L | apache-2.0 | BytedTsinghua-SIA | 32 | 20k |
| clouditera/secgpt | finetune | — | 14.8B | L | apache-2.0 | clouditera | 107 | 14.9k |
| lmstudio-community/Qwen2.5-14B-Instruct-MLX-8bit | quantized | MLX | 14.8B | L | apache-2.0 | lmstudio-community | 0 | 19.4k |
| mradermacher/Qwen2.5-14B-Instruct-i1-GGUF | quantized | GGUF | — | — | apache-2.0 | mradermacher | 2 | 16.6k |
| NCSOFT/VARCO-VISION-14B-HF | finetune | — | 15.2B | L | cc-by-nc-4.0 | NCSOFT | 30 | 15k |
| Goekdeniz-Guelmez/Josiefied-Qwen2.5-14B-Instruct-abliterated-v4 | finetune | — | 14.8B | L | apache-2.0 | Goekdeniz-Guelmez | 20 | 15.2k |
| second-state/Qwen2.5-14B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | second-state | 0 | 14.1k |
| Sao10K/14B-Qwen2.5-Kunou-v1 | finetune | — | 14.8B | L | other | Sao10K | 36 | 11.2k |
| simplescaling/s1.1-14B | finetune | — | 14.8B | L | — | simplescaling | 4 | 12.2k |
| gaianet/Qwen2.5-14B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | gaianet | 0 | 10.9k |
| QuantFactory/Qwen2.5-14B-Gutenberg-1e-Delta-GGUF | quantized | GGUF | — | — | apache-2.0 | QuantFactory | 2 | 9.9k |
| QuantFactory/Qwen2.5-14B-Gutenberg-Instruct-Slerpeno-GGUF | merge | GGUF | — | — | apache-2.0 | QuantFactory | 2 | 9.3k |
| MaziyarPanahi/Qwen2.5-14B-Instruct-GGUF | quantized | GGUF | — | — | — | MaziyarPanahi | 2 | 8.4k |
| Mungert/OpenReasoning-Nemotron-14B-GGUF | quantized | GGUF | — | — | cc-by-4.0 | Mungert | 3 | 8.1k |
| Orion-zhen/Qwen2.5-14B-Instruct-Uncensored | finetune | — | 14.8B | L | gpl-3.0 | Orion-zhen | 27 | 6.9k |
| nvidia/OpenCodeReasoning-Nemotron-14B | finetune | — | 14.8B | L | apache-2.0 | nvidia | 24 | 7k |
| Date | Commit | SHA |
|---|---|---|
| 2024-09-25 | Update metadata: Add library_name to Transformers (#4) | cf98f3b |
| 2024-09-19 | Upload ./tokenizer_config.json with huggingface_hub | 502e5d8 |
| 2024-09-18 | update README & LICENSE | f55224c |
| 2024-09-18 | update config.json | 76ca6ba |
| 2024-09-18 | Update tokenizer_config.json | e648520 |
| 2024-09-17 | Update README.md | e4e2ea5 |
| 2024-09-17 | Update README.md | e5147e7 |
| 2024-09-17 | Create README.md | e40aee8 |
| 2024-09-16 | Upload folder using huggingface_hub | 336cd7f |
| 2024-09-16 | initial commit | 458a599 |
Hosted endpoints observed for Qwen2.5-14B-Instruct, sorted by input price (USD per million tokens). Each row names its observer and date; prices change daily, so check the source before quoting.
| Provider | Model ref | Context | Quant | $/M in | $/M out | tok/s | TTFT | Tools | Status | Observed |
|---|---|---|---|---|---|---|---|---|---|---|
| Featherless hf | Qwen/Qwen2.5-14B-Instruct | — | — | — | — | — | — | — | live | 2026-10-25 |
Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.
arXiv links found in the model card and its tags. "Peer-reviewed" is shown only with evidence: a published venue in OpenAlex (measured) or the authors' own journal reference on arXiv (reported).
Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:35 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:01 UTC · Last verified 2026-10-25
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.
