Model Follower.

Up to date

Rebuilt
2026-10-25 06:54 UTC
Release radar
117 on watch
Hub snapshot
2026-10-25
Re-checked
daily · every number sourced and dated

Models › Qwen3 family

Qwen3-Coder-30B-A3B-Instruct

Hugging Face ↗ Follow changes

Qwen Apache 2.0 out 15 months ago

30.5B parameters, 3B active (8 of 128 experts per token), text in, text out, 256K tokens of context. 3 providers serve it from $0.070 per million input tokens.

Verified 2026-10-25 Rank #45 in LLMs · score 108.7 method v1 Sources 4 Downloads 30d 700k · #140 of 619 tracked

Changed this week: 3 new quantized builds on the Hub (180 in all).2026-10-24 · 1 provider no longer listed: Scaleway.2026-10-23

Can I run it?18GB at INT4Estimatefits a 24 GB cardmethodCommunity: Qwen3 Coder 30B A3B Instruct — Parameters, Context, Benchmarks & Pricing | AI Atlas →
What does it cost?$0.070/1M incheapest of 2SiliconFlowfp82026-10-24
Can I use it?Apache 2.0Permissive; patent grantlicence file2026-10-24
Does it work?49.67on SWE-bench#1 of 4 measuredleaderboard2025-09-01
Do people use it?700kdownloads, 30 days#140 of 619 we track · top 23%+14 likes, downloads +50% this week1,306 likesHugging Face2026-10-24

Worth reading

Picked by rule from 19 tracked pieces: the lab's own post, the most shared, a field report, the latest.

Most discussed · 490 HN pts

GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

OpenAI released gpt-oss-120b and gpt-oss-20b, their first open-weight models since GPT-2 in 2019.

  • A 2025 research paper confirms Dropout results in worse downstream performance in single-epoch regimes.
magazine.sebastianraschka.com 2025-08-10 490 HN pts · 97 comments 5.2k words about the Qwen3 family family

19 best of 19 tracked

The 3 picks above, plus these 16. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.

  1. ArticleQwen3-Coder-Nextqwen.ai · 2026-02-03 · 735 HN pts · official
  2. ArticleQwen3-Max-Thinkingqwen.ai · 2026-01-26 · official
  3. ArticleQwen3-TTS family is now open sourced: Voice design, clone, and generationqwen.ai · 2026-01-22 · official
  4. ArticleQwen3: Think deeper, act fasterqwenlm.github.io · 2025-04-28 · 869 HN pts · official
  5. ArticleQwen3-Coder: Agentic coding in the worldqwenlm.github.io · 2025-07-22 · 765 HN pts · official
  6. ArticleQwen3-Nextqwen.ai · 2025-09-12 · official
  7. ArticleQwen3-VLqwen.ai · 2025-09-23 · official
  8. CodeQwen3-Omni: Native Omni AI model for text, image and videogithub.com · 2025-09-22
  9. ArticleQwen3 Coder 30B A3B Instruct - API Pricing & Benchmarks | OpenRouteropenrouter.ai · undated
  10. CodeREADME.mdgithub.com · undated
  11. ArticleQwen3 Coder 30B A3B Instruct — Parameters, Context, Benchmarks & Pricing | AI Atlasai-atlas.co · undated
  12. ArticleQwen3 Coder 30B A3B Instruct: benchmarks & pricing — ModelCap #106modelcap.ai · undated
  13. ArticleQwen: Qwen3 Coder 30B A3B Instruct - AI Model Details & B...benchable.ai · undated
  14. Articleqwen--qwen3-coder-30b-a3b-... | Model Catalog | Microsoft Foundryai.azure.com · undated
  15. ArticleQwen3 Coder 30B A3B Instruct: specs, context window and benchmarks | FastMetalfastmetal.ai · undated
  16. ArticleLocal AI agent in 44 lines of code powered by Qwen3 30B A3B Instructrentry.co · 2025-08-15

Where to run it

Live hosted endpoints, cheapest first, USD per million tokens. Ties break on output price, then provider name.

Prices for the same weights vary 2×

2 providers with a live price, from $0.070 to $0.15 per million input tokens; the median is $0.11. One dot per provider, cheapest in teal.

$0.010$0.10$1.00$10.00SiliconFlow: $0.070 in / $0.28 out per 1M · observed 2026-10-24 · https://openrouter.ai/qwen/qwen3-coder-30b-a3b-instructAmazon Bedrock: $0.15 in / $0.60 out per 1M · observed 2026-10-24 · https://openrouter.ai/qwen/qwen3-coder-30b-a3b-instructcheapest $0.070median $0.11
$0.010$0.10$1.00$10.00SiliconFlow: $0.070 in / $0.28 out per 1M · observed 2026-10-24 · https://openrouter.ai/qwen/qwen3-coder-30b-a3b-instructAmazon Bedrock: $0.15 in / $0.60 out per 1M · observed 2026-10-24 · https://openrouter.ai/qwen/qwen3-coder-30b-a3b-instructcheapest $0.070 · median $0.11 (white tick)
#Cheapest 3In /1MOut /1MContextQuantSeen
1 SiliconFlow $0.070 $0.28 256K fp8 Oct 24
2 Amazon Bedrock $0.15 $0.60 256K unknown Oct 24
3 Featherless — — — — Oct 25

All 3 providers →

Where it's used

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 3 API providers; heavily run locally (180 quantized builds); a big base for builders (167 fine-tunes, adapters and merges).

  • As a service3Amazon Bedrock, Featherless, SiliconFlow
  • On own hardware180quantized builds
  • Built on16798 fine-tunes · 63 adapters · 6 merges
  • In apps100+Hugging Face Spaces

Top Spaces running it: enzostvs/deepsite ♥ 16,604 · AiSudo/Qwen-Image-to-LoRA ♥ 344 · infinite-dataset-hub/infinite-dataset-hub ♥ 297 · AiSudo/Z-Image-to-LoRA ♥ 59

Will it fit?

Weights only, an estimate: leave room for the KV cache. Method

Fits a 64 GB Mac (unified memory) at INT4

About 18 GB at INT4, 37 GB at FP8, 73 GB at BF16: weights only, so leave room for the KV cache. Every expert must be resident, so total parameters count.

16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 18 GB of weights (estimate; KV cache and activations extra)INT418 GBFP8: about 37 GB of weights (estimate; KV cache and activations extra)FP837 GBBF16: about 73 GB of weights (estimate; KV cache and activations extra)BF1673 GB
16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 18 GB of weights (estimate; KV cache and activations extra)INT418 GBFP8: about 37 GB of weights (estimate; KV cache and activations extra)FP837 GBBF16: about 73 GB of weights (estimate; KV cache and activations extra)BF1673 GB

Estimate params × bytes per param × 1.2 · how we estimate

How it scores

Third-party measurements only; author-reported numbers stay in Reference, marked as such.

#1 of 4 on SWE-bench so far

49.67. Each tick is a model on this site with a measured SWE-bench row; author-reported numbers stay in Reference. Percent resolved. agent: EntroPO + R2E

deepseek-ai/DeepSeek-V3: 36.67 · observed 2026-10-25 · https://www.swebench.com/deepseek-ai/DeepSeek-V3: 30.67 · observed 2026-10-25 · https://www.swebench.com/010049.67 · #1 of 4
deepseek-ai/DeepSeek-V3: 36.67 · observed 2026-10-25 · https://www.swebench.com/deepseek-ai/DeepSeek-V3: 30.67 · observed 2026-10-25 · https://www.swebench.com/010049.67 · #1 of 4

Source · leaderboard · 2025-09-01

The story so far

Weights, every dated article, and repo commits on one rail.

Most coverage came in the first two weeks

Teal is the weights (2025-07-31), amber triangles are 6 dated articles (each links to it), grey ticks are repo commits.

MayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOct2025-12-03: 1 repo commit2025-08-21: 1 repo commit2025-08-07: 1 repo commit2025-08-01: 1 repo commit2025-07-31: 9 repo commitsQwen3-Coder-Next · 2026-02-03Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers · 2026-02-28Qwen3: Think deeper, act faster · 2025-04-28Qwen3-Coder: Agentic coding in the world · 2025-07-22GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2 · 2025-08-10Qwen3-Coder-30B-A3B-Instruct — model deep-dive · Tokonomix · 2026-05-05Weights 2025-07-31 · 2025-07-31Weights 2025-07-31now · 3 providers
MayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOct2025-12-03: 1 repo commit2025-08-21: 1 repo commit2025-08-07: 1 repo commit2025-08-01: 1 repo commit2025-07-31: 9 repo commitsQwen3-Coder-Next · 2026-02-03Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers · 2026-02-28Qwen3: Think deeper, act faster · 2025-04-28Qwen3-Coder: Agentic coding in the world · 2025-07-22GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2 · 2025-08-10Qwen3-Coder-30B-A3B-Instruct — model deep-dive · Tokonomix · 2026-05-05Weights 2025-07-31 · 2025-07-31Weights 2025-07-31now · 3 providers

What people built on it

347 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.

347 derivatives on the Hub, most of them quantized

From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.

180 quantized (Hugging Face model tree)98 fine-tunes (Hugging Face model tree)6 merges (Hugging Face model tree)63 adapters (Hugging Face model tree)180 quantized98 fine-tunes6 merges63 adapters
180 quantized (Hugging Face model tree)98 fine-tunes (Hugging Face model tree)6 merges (Hugging Face model tree)63 adapters (Hugging Face model tree)180 quantized98 fine-tunes6 merges63 adapters
GGUF unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF for llama.cpp, Ollama, LM Studio 28.9M downloads MLX lmstudio-community/Qwen3-Coder-30B-A3B-Instruct-MLX-4bit for Apple silicon 3M downloads AWQ cyankiwi/Qwen3-Coder-30B-A3B-Instruct-AWQ-4bit for vLLM, SGLang 5M downloads

All variants →

Reference

Everything else, folded: the full tables, every source.

Benchmarks

5 rows

Benchmarks 5 rows

measured third-party leaderboard row (5)author-reported number from the model card or eval filesNo confidence intervals unless the source gave them; settings differ between rows, so compare within a benchmark, not across. Date = the benchmark's own date when it gives one, else when we read it.
BenchmarkValueSetting / notesReported byDate
measured SWE-bench · Write code with agents49.67 percent resolvedagent: EntroPO + R2Eleaderboard2025-09-01
measured SWE-bench · Write code with agents45 percent resolvedagent: EntroPO + R2Eleaderboard2025-09-01
measured SWE-bench Verified · Write code with agents60.4 percent resolvedagent: EntroPO + R2Eleaderboard2025-09-01
measured SWE-bench Verified · Write code with agents52.2 percent resolvedagent: EntroPO + R2Eleaderboard2025-09-01
measured SWE-bench Verified · Write code with agents51.6 percent resolvedagent: OpenHandsleaderboard2025-08-05

What the card says

1 cited sentence

Summary generated, cited

It features a native context length of 262,144 tokens and is released under the Apache 2.0 license.[1][2]

  1. “Context Length: **262,144 natively**.”
  2. license

Generated 2026-10-07 14:59 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.

Use-case evidence

4 use cases

What it is good at 4 use cases

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.

Facts

24 fields

Listed in LLMs · MoE

Facts

Pipeline tagtext-generation
Librarytransformers
ArchitecturesQwen3MoeForCausalLM
Model typeqwen3_moe
Dtypebfloat16
Params by dtypeBF16: 30,532,122,624
Size bandL
MoEyes
Experts128
Experts per token8
Context length262,144 (config.json max_position_embeddings)
PapersarXiv:2505.09388
LicenseApache 2.0 (apache-2.0)
Hub snapshot2026-10-25

Servability

vLLM · SGLang

Servability

vLLM: supported · Qwen3MoeForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · Qwen3MoeForCausalLM · 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.

Hardware fit

3 estimates

Hardware fit ESTIMATE

Weights-only memory for Qwen3-Coder-30B-A3B-Instruct: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.

BF16 (2 bytes/param)73 GB · Single high-end GPU (<80 GB)
FP8 (1 byte/param)37 GB · Single high-end GPU (<80 GB)
INT4 (0.5 bytes/param)18 GB · Single GPU (<24 GB)

Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra. MoE: all experts must be resident, so total parameters count.

All variants

40 variants

Lineage

Original model Family: Qwen/Qwen3 347 derivatives on the Hub

quantized: 180finetune: 98adapter: 63merge: 6

Top variants (40 shown, ranked by downloads)

Relation
Quant
Size
40 rows
Variant Relation Quant Params Band License Author Likes Downloads
unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF quantized GGUF — — apache-2.0 unsloth 1.2k 28.9M
cyankiwi/Qwen3-Coder-30B-A3B-Instruct-AWQ-4bit quantized AWQ 5.31B M apache-2.0 cyankiwi 67 5M
lmstudio-community/Qwen3-Coder-30B-A3B-Instruct-MLX-4bit quantized MLX 30.5B L apache-2.0 lmstudio-community 40 3M
lmstudio-community/Qwen3-Coder-30B-A3B-Instruct-MLX-8bit quantized MLX 30.5B L apache-2.0 lmstudio-community 26 2.8M
lmstudio-community/Qwen3-Coder-30B-A3B-Instruct-MLX-6bit quantized MLX 30.5B L apache-2.0 lmstudio-community 8 2.7M
lmstudio-community/Qwen3-Coder-30B-A3B-Instruct-MLX-5bit quantized MLX 30.5B L apache-2.0 lmstudio-community 10 2.7M
QuantTrio/Qwen3-Coder-30B-A3B-Instruct-AWQ quantized AWQ 30.5B L apache-2.0 QuantTrio 11 2.6M
lmstudio-community/Qwen3-Coder-30B-A3B-Instruct-GGUF quantized GGUF — — — lmstudio-community 44 952.3k
ggml-org/Qwen3-Coder-30B-A3B-Instruct-Q8_0-GGUF quantized Q8_0 — — apache-2.0 ggml-org 12 516.6k
stelterlab/Qwen3-Coder-30B-A3B-Instruct-AWQ quantized AWQ 30.8B L apache-2.0 stelterlab 9 447k
btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit quantized gptq 30.5B L apache-2.0 btbtyler09 5 403.7k
nm-testing/Qwen3-Coder-30B-A3B-Instruct-W4A16-awq quantized -W4A16 4.61B M — nm-testing 4 349.3k
unsloth/Qwen3-Coder-30B-A3B-Instruct-1M-GGUF quantized GGUF — — apache-2.0 unsloth 165 299.9k
QuantTrio/Qwen3-Coder-30B-A3B-Instruct-GPTQ-Int8 quantized GPTQ 30.5B L apache-2.0 QuantTrio 8 289.4k
unsloth/Qwen3-Coder-30B-A3B-Instruct finetune — 30.5B L apache-2.0 unsloth 29 217.4k
NVFP4/Qwen3-Coder-30B-A3B-Instruct-FP4 quantized — 15.6B L apache-2.0 NVFP4 38 216.1k
btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit quantized gptq 30.5B L apache-2.0 btbtyler09 4 144.8k
mlx-community/Qwen3-Coder-30B-A3B-Instruct-4bit quantized -4bit 30.5B L apache-2.0 mlx-community 51 95.7k
cerebras/Qwen3-Coder-REAP-25B-A3B finetune — 24.9B L apache-2.0 cerebras 91 68.2k
huihui-ai/Huihui-Qwen3-Coder-30B-A3B-Instruct-abliterated finetune — 30.5B L apache-2.0 huihui-ai 69 61k
ig1/Qwen3-Coder-30B-A3B-Instruct-NVFP4 quantized NVFP4 17.5B L apache-2.0 ig1 1 59.2k
mradermacher/Qwen3-Coder-30B-A3B-Instruct-i1-GGUF quantized GGUF — — apache-2.0 mradermacher 11 43.8k
byteshape/Qwen3-Coder-30B-A3B-Instruct-GGUF quantized GGUF — — apache-2.0 byteshape 41 30.8k
n00b001/Qwen3-Coder-30B-A3B-Instruct-Q4_K_M-GGUF quantized GGUF — — apache-2.0 n00b001 5 29.8k
AtomicChat/Qwen3-Coder-30B-A3B-GGUF quantized GGUF — — apache-2.0 AtomicChat 4 28.5k
mudler/Qwen3-Coder-30B-APEX-GGUF quantized GGUF — — apache-2.0 mudler 16 27.3k
dark-side-of-the-code/Qwen3-Coder-30B-A3B-Instruct-AWQ quantized AWQ 4.61B M apache-2.0 dark-side-of-the-code 1 24.8k
CISCai/Qwen3-Coder-30B-A3B-Instruct-SOTA-GGUF quantized GGUF — — apache-2.0 CISCai 0 22.7k
ramblingpolymath/Qwen3-Coder-30B-A3B-Instruct-W8A8 quantized -W8A8 30.6B L apache-2.0 ramblingpolymath 3 22.4k
tensorblock/Qwen_Qwen3-Coder-30B-A3B-Instruct-GGUF quantized GGUF — — apache-2.0 tensorblock 3 21.3k
geoffmunn/Qwen3-Coder-30B-A3B-Instruct-f16 quantized GGUF — — apache-2.0 geoffmunn 4 20.2k
giladgd/Qwen3-Coder-30B-A3B-Instruct-Q4_K_M-GGUF quantized GGUF — — apache-2.0 giladgd 5 20k
second-state/Qwen3-Coder-30B-A3B-Instruct-GGUF quantized GGUF — — apache-2.0 second-state 0 19.5k
mradermacher/Qwen3-Coder-30B-A3B-Instruct-GGUF quantized GGUF — — apache-2.0 mradermacher 2 18.4k
Intel/Qwen3-Coder-30B-A3B-Instruct-int4-AutoRound quantized int4 635M S apache-2.0 Intel 9 17.8k
tenzro/daml-coder finetune — 30.5B L apache-2.0 tenzro 0 17.4k
mlx-community/Qwen3-Coder-30B-A3B-Instruct-4bit-dwq-v2 quantized -4bit 30.5B L apache-2.0 mlx-community 11 16.3k
AIDXteam/Qwen3-Coder-30B-A3B-Instruct-AWQ quantized AWQ 30.8B L apache-2.0 AIDXteam 1 16.5k
OPENZEKA/Qwen3-Coder-30B-A3B-Instruct-NVFP4 quantized NVFP4 15.6B L apache-2.0 OPENZEKA 0 16.2k
continuum-ai/qwen3-coder-30b-a3b-compacted-19b-256k quantized GGUF — — apache-2.0 continuum-ai 16 14.7k

Repository changelog

13 commits
changed in the last 7 daysProviders dropped: 2026-10-23Other: 2026-10-24all changes

Repository changelog 13 commits

DateCommitSHA
2025-12-03 Update config.json b2cff64
2025-08-21 Update chat_template and tool_parser (#22) 573fa39
2025-08-07 Update chat_template in tokenizer_config.json (#14) 8110d98
2025-08-01 Update README.md d4add6d
2025-07-31 Update README.md c80c45f
2025-07-31 Update README.md 08f5d08
2025-07-31 Create LICENSE 62dd7e6
2025-07-31 Update README.md ac707c8
2025-07-31 Update README.md 82caff9
2025-07-31 Update README.md d38ea9e
2025-07-31 Delete .md5sum_check.txt 1872883
2025-07-31 Upload folder using huggingface_hub 703eb35
2025-07-31 initial commit 87fefa3

All providers

3 providers · 3 endpoints

Where to run 3 providers · 3 endpoints

Hosted endpoints observed for Qwen3-Coder-30B-A3B-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. Cheapest observed input price: $0.07/M.

ProviderModel refContextQuant$/M in$/M outtok/sTTFTToolsStatusObserved
SiliconFlow
openrouter
 qwen/qwen3-coder-30b-a3b-instruct 256K fp8 $0.07 $0.28 — — yes live 2026-10-24
Amazon Bedrock
openrouter
 qwen/qwen3-coder-30b-a3b-instruct 256K unknown $0.15 $0.60 — — yes live 2026-10-24
Featherless
hf
 Qwen/Qwen3-Coder-30B-A3B-Instruct — — — — — — — live 2026-10-25

Articles and coverage

19 pieces

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.

Official3

Qwen3-Coder-Next
qwen.ai2026-02-03officialauthorityHN front pageabout the Qwen3 familyHN 735 pts · 429 comments674 words
Qwen3-Coder-Next is an open-weight language model designed for coding agents and local development. The model achieves over 70% on SWE-Bench Verified using the SWE-Agent scaffold while maintaining a 3B active footprint. This small active footprint achieves SWE-Bench-Pro performance comparable to models with 10x-20x more active parameters.written, reviewed, cited
  • The model is built on top of Qwen3-Next-80B-A3B-Base, which adopts a novel architecture with hybrid attention and MoE.
  • Training includes continued pretraining on code- and agent-centric data alongside supervised fine-tuning on high-quality agent trajectories.
  • The model emphasizes long-horizon reasoning, tool usage, and recovery from execution failures for real-world coding agents.
Qwen3: Think deeper, act faster
qwenlm.github.io2025-04-28officialauthorityHN front pageabout the Qwen3 familyHN 869 pts · 388 comments2,139 words
Qwen3 introduces Hybrid Thinking Modes, allowing users to toggle between step-by-step reasoning and rapid responses via computational budget control. The open-weight MoE model Qwen3-235B-A22B achieves performance similar to dense models while using only 10% of the active parameters. Pre-training utilized approximately 36 trillion tokens across 119 languages, with deployment supported by SGLang and vLLM.written, reviewed, cited
  • Qwen3-8B and Qwen3-14B models do not use tied embeddings, unlike smaller variants.
  • The pre-training pipeline includes a final stage extending context length to 32K tokens.
Qwen3-Coder: Agentic coding in the world
qwenlm.github.io2025-07-22officialauthorityHN front pageabout the Qwen3 familyHN 765 pts · 366 comments958 words

Analysis7

Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers
venturebeat.com2026-02-28HN front pageabout the Qwen3 familyHN 461 pts · 272 comments
GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2
magazine.sebastianraschka.com2025-08-10authorityHN front pageabout the Qwen3 familyHN 490 pts · 97 comments5,195 words
OpenAI released gpt-oss-120b and gpt-oss-20b, their first open-weight models since GPT-2 in 2019. The 20B model can run on a consumer GPU with up to 16 GB of RAM, and the 120B model can run on a single H100 with 80 GB of RAM or newer hardware. The models utilize MXFP4 optimization to fit onto single GPUs.written, reviewed, cited
  • A 2025 research paper confirms Dropout results in worse downstream performance in single-epoch regimes.
Qwen3 Coder 30B A3B Instruct - API Pricing & Benchmarks | OpenRouter
openrouter.aiauthority
README.md
github.comauthorityabout the Qwen3 family
Qwen3-Coder-30B-A3B-Instruct — model deep-dive · Tokonomix
tokonomix.ai2026-05-05
Qwen3 Coder 30B A3B Instruct — Parameters, Context, Benchmarks & Pricing | AI Atlas
ai-atlas.co
Qwen3 Coder 30B A3B Instruct: benchmarks & pricing — ModelCap #106
modelcap.ai
All coverage tracked for this model (19) — append-only, nothing is ever removed
TitleSourcePublishedFirst seenSignals
Qwen3 Coder 30B A3B Instruct - API Pricing & Benchmarks | OpenRouteropenrouter.ai—2026-10-07exa
README.mdgithub.com—2026-10-07exa
Qwen3 Coder 30B A3B Instruct — Parameters, Context, Benchmarks & Pricing | AI Atlasai-atlas.co—2026-10-07exa
Qwen3 Coder 30B A3B Instruct: benchmarks & pricing — ModelCap #106modelcap.ai—2026-10-07exa
Qwen: Qwen3 Coder 30B A3B Instruct - AI Model Details & B...benchable.ai—2026-10-07exa
qwen--qwen3-coder-30b-a3b-... | Model Catalog | Microsoft Foundryai.azure.com—2026-10-07exa
Qwen3 Coder 30B A3B Instruct: specs, context window and benchmarks | FastMetalfastmetal.ai—2026-10-07exa
Qwen3-Coder-30B-A3B-Instruct — model deep-dive · Tokonomixtokonomix.ai2026-05-052026-10-07exa
Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computersventurebeat.com2026-02-282026-10-07hn
Qwen3-Coder-Nextqwen.ai official2026-02-032026-10-07hn
Qwen3-Max-Thinkingqwen.ai official2026-01-262026-10-07hn
Qwen3-TTS family is now open sourced: Voice design, clone, and generationqwen.ai official2026-01-222026-10-07hn
Qwen3-VLqwen.ai official2025-09-232026-10-07hn
Qwen3-Omni: Native Omni AI model for text, image and videogithub.com2025-09-222026-10-07hn
Qwen3-Nextqwen.ai official2025-09-122026-10-07hn
Local AI agent in 44 lines of code powered by Qwen3 30B A3B Instructrentry.co2025-08-152026-10-07hn
GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2magazine.sebastianraschka.com2025-08-102026-10-07hn
Qwen3-Coder: Agentic coding in the worldqwenlm.github.io official2025-07-222026-10-07hn
Qwen3: Think deeper, act fasterqwenlm.github.io official2025-04-282026-10-07hn

Papers

1 paper

Papers 1

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).

Similar models

3 models

Similar models

  • GLM-4.7-Flash · 31.2B
    same size band (L), also MoE, same task (text-generation), more providers, permissive license
  • North-Mini-Code-1.0 · 30.5B
    same size band (L), also MoE, same task (text-generation), fewer providers, permissive license
  • Tongyi-DeepResearch-30B-A3B · 30.5B
    same size band (L), also MoE, same task (text-generation), permissive license

Sources

4 sources

Sources

Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:32 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 14:59 UTC · Last verified 2026-10-25

Follow Qwen3-Coder-30B-A3B-Instruct

Watch this model

Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.