Up to date
deepseek-ai MIT out 21 months ago
1.78B parameters, text in, text out, 128K tokens of context. 1 provider serves it.
Verified 2026-10-25 Rank #85 in LLMs · score 102.1 method v1 Sources 3 Downloads 30d 1.3M · #95 of 619 tracked
Changed this week: 1 new quantized build on the Hub (277 in all).2026-10-24
Picked by rule from 33 tracked pieces: the lab's own post, the most shared, a field report, the latest.
Zyphra released ZAYA1-8B, a model that matches DeepSeek-R1 on math benchmarks while using less than 1 billion active 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 (277 quantized builds); a big base for builders (1,071 fine-tunes, adapters and merges).
Top Spaces running it: pliny-the-prompter/obliteratus ♥ 621 · hesamation/primer-llm-embedding ♥ 374 · KBaba7/Quant ♥ 51 · FallnAI/Quantize-HF-Models ♥ 19
Weights only, an estimate: leave room for the KV cache. Method
About 1.1 GB at INT4, 2.1 GB at FP8, 4.3 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 (2025-01-20), amber triangles are 10 dated articles (each links to it), grey ticks are repo commits.
1,348 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.
| Benchmark | Value | Setting / notes | Reported by | Date |
|---|---|---|---|---|
| author-reported AIME 2024 · Reason through hard problems | 28.9 pass@1 | DeepSeek-R1-Distill-Qwen-1.5B | model author | 2026-10-07 |
| author-reported AIME 2024 · Reason through hard problems | 52.7 other | DeepSeek-R1-Distill-Qwen-1.5B | model author | 2026-10-07 |
| author-reported Codeforces · Write code with agents | 954 rating | DeepSeek-R1-Distill-Qwen-1.5B | model author | 2026-10-07 |
| author-reported GPQA Diamond · Reason through hard problems | 33.8 pass@1 | DeepSeek-R1-Distill-Qwen-1.5B | model author | 2026-10-07 |
| author-reported LiveCodeBench · Write code with agents | 16.9 pass@1 | DeepSeek-R1-Distill-Qwen-1.5B | model author | 2026-10-07 |
| author-reported MATH-500 · Reason through hard problems | 83.9 pass@1 | DeepSeek-R1-Distill-Qwen-1.5B | model author | 2026-10-07 |
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: 1,777,088,000 |
| Size band | S |
| MoE | no |
| Context length | 131,072 (config.json max_position_embeddings) |
| Papers | arXiv:2501.12948 |
| License | MIT (mit) |
| 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 DeepSeek-R1-Distill-Qwen-1.5B: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 4.3 GB · Single GPU (<24 GB) |
|---|---|
| FP8 (1 byte/param) | 2.1 GB · Edge (<4 GB) |
| INT4 (0.5 bytes/param) | 1.1 GB · Edge (<4 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
quantized: 277finetune: 704adapter: 340merge: 27
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| unsloth/DeepSeek-R1-Distill-Qwen-1.5B-GGUF | quantized | GGUF | — | — | apache-2.0 | unsloth | 165 | 1.7M |
| litert-community/DeepSeek-R1-Distill-Qwen-1.5B | finetune | — | — | — | mit | litert-community | 56 | 819.7k |
| agentica-org/DeepScaleR-1.5B-Preview | finetune | — | 1.78B | S | mit | agentica-org | 584 | 752.6k |
| bartowski/DeepSeek-R1-Distill-Qwen-1.5B-GGUF | quantized | GGUF | — | — | — | bartowski | 91 | 637k |
| unsloth/DeepSeek-R1-Distill-Qwen-1.5B-unsloth-bnb-4bit | quantized | -bnb | 1.81B | S | apache-2.0 | unsloth | 17 | 354.3k |
| lmstudio-community/DeepSeek-R1-Distill-Qwen-1.5B-GGUF | quantized | GGUF | — | — | — | lmstudio-community | 16 | 223.8k |
| mlx-community/deepseek-r1-distill-qwen-1.5b | finetune | — | 1.78B | S | — | mlx-community | 24 | 209.1k |
| unsloth/DeepSeek-R1-Distill-Qwen-1.5B-bnb-4bit | quantized | -bnb | 1.82B | S | apache-2.0 | unsloth | 11 | 191.9k |
| nvidia/DLER-R1-1.5B-Research | finetune | — | 1.78B | S | — | nvidia | 21 | 128.9k |
| unsloth/DeepSeek-R1-Distill-Qwen-1.5B | finetune | — | 1.78B | S | mit | unsloth | 16 | 104.5k |
| salakash/SamKash-Tolstoy | adapter | — | — | — | apache-2.0 | salakash | 758 | 42.5k |
| onnx-community/DeepSeek-R1-Distill-Qwen-1.5B-ONNX | quantized | — | — | — | — | onnx-community | 61 | 75.8k |
| mlx-community/DeepSeek-R1-Distill-Qwen-1.5B-8bit | quantized | -8bit | 1.78B | S | — | mlx-community | 3 | 77.3k |
| mlx-community/DeepSeek-R1-Distill-Qwen-1.5B-4bit | quantized | -4bit | 1.78B | S | — | mlx-community | 6 | 72.5k |
| RedHatAI/DeepSeek-R1-Distill-Qwen-1.5B-quantized.w8a8 | quantized | -quant | 1.78B | S | mit | RedHatAI | 2 | 50.6k |
| Zyphra/ZR1-1.5B | finetune | — | 1.78B | S | mit | Zyphra | 77 | 44.5k |
| agentica-org/DeepCoder-1.5B-Preview | finetune | — | 1.78B | S | mit | agentica-org | 76 | 34.1k |
| Sweaterdog/Andy-3.5 | quantized | GGUF | — | — | apache-2.0 | Sweaterdog | 6 | 36k |
| knoveleng/Open-RS3 | finetune | — | 1.78B | S | mit | knoveleng | 21 | 33.6k |
| ggml-org/DeepSeek-R1-Distill-Qwen-1.5B-Q4_0-GGUF | quantized | Q4_0 | — | — | — | ggml-org | 5 | 34.4k |
| OpenVINO/DeepSeek-R1-Distill-Qwen-1.5B-int4-ov | quantized | int4 | — | — | mit | OpenVINO | 2 | 30.9k |
| hbx/JustRL-DeepSeek-1.5B | finetune | — | 1.78B | S | apache-2.0 | hbx | 12 | 27.9k |
| Brianpuz/DeepSeek-R1-Distill-Qwen-1.5B-Q4_0-Q2_K-GGUF | quantized | Q4_0 | — | — | mit | Brianpuz | 0 | 22.8k |
| Vinnnf/Thinkless-1.5B-Warmup | finetune | — | 1.78B | S | apache-2.0 | Vinnnf | 4 | 19.9k |
| DevQuasar-7/deepseek-ai.DeepSeek-R1-Distill-Qwen-1.5B-GGUF | quantized | GGUF | — | — | — | DevQuasar-7 | 2 | 18.9k |
| GenPRM/GenPRM-1.5B | finetune | — | 1.78B | S | mit | GenPRM | 2 | 17.8k |
| second-state/DeepSeek-R1-Distill-Qwen-1.5B-GGUF | quantized | GGUF | — | — | mit | second-state | 0 | 16.1k |
| mlx-community/DeepSeek-R1-Distill-Qwen-1.5B-6bit | quantized | — | 1.78B | S | — | mlx-community | 0 | 16.1k |
| huihui-ai/DeepSeek-R1-Distill-Qwen-1.5B-abliterated | finetune | — | 1.78B | S | — | huihui-ai | 7 | 15.6k |
| NexaAI/DeepSeek-R1-Distill-Qwen-1.5B-NexaQuant | quantized | GGUF | — | — | apache-2.0 | NexaAI | 92 | 11.2k |
| 0xfaskety/task-1-microsoft-Phi-3-mini-4k-instruct | adapter | — | — | — | — | 0xfaskety | 1 | 15.3k |
| knoveleng/Open-RS2 | finetune | — | 1.78B | S | mit | knoveleng | 1 | 13.6k |
| ThomasBaruzier/DeepSeek-R1-ReDistill-Qwen-1.5B-v1.1-GGUF | quantized | GGUF | — | — | mit | ThomasBaruzier | 0 | 13.5k |
| knoveleng/Open-RS1 | finetune | — | 1.78B | S | mit | knoveleng | 4 | 13.1k |
| QuantFactory/DeepScaleR-1.5B-Preview-GGUF | quantized | GGUF | — | — | mit | QuantFactory | 2 | 12.5k |
| mlx-community/DeepSeek-R1-Distill-Qwen-1.5B-3bit | quantized | — | 1.78B | S | — | mlx-community | 1 | 12.3k |
| Mungert/DeepCoder-1.5B-Preview-GGUF | quantized | GGUF | — | — | mit | Mungert | 4 | 11.9k |
| Mungert/Nemotron-Research-Reasoning-Qwen-1.5B-GGUF | quantized | GGUF | — | — | cc-by-nc-4.0 | Mungert | 2 | 10.6k |
| hdnh2006/DeepSeek-R1-Distill-Qwen-1.5B-GGUF | quantized | GGUF | — | — | apache-2.0 | hdnh2006 | 3 | 10.5k |
| zwhe99/DeepMath-1.5B | finetune | — | 1.78B | S | mit | zwhe99 | 5 | 10.2k |
| Date | Commit | SHA |
|---|---|---|
| 2025-02-24 | Update bib info | ad9f0ae |
| 2025-02-09 | Update tokenizer_config.json | 530ca3e |
| 2025-02-01 | Update README.md | 6393b75 |
| 2025-01-26 | Update README.md | 0092dee |
| 2025-01-23 | Update README.md | 5d85599 |
| 2025-01-22 | Update tokenizer_config.json | 164af3f |
| 2025-01-22 | Update tokenizer_config.json | e937840 |
| 2025-01-21 | Add generation_config.json | 73d63fa |
| 2025-01-21 | add library tag for better code snippets and tags (#1) | aa76ddc |
| 2025-01-20 | Release DeepSeek-R1 | 80da49e |
| 2025-01-20 | Add files using upload-large-folder tool | e164738 |
| 2025-01-20 | initial commit | 8f1dfda |
Hosted endpoints observed for DeepSeek-R1-Distill-Qwen-1.5B, 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 | deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B | — | — | — | — | — | — | — | 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 21:33 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:00 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.
