Up to date
Qwen Apache 2.0 out 14 months ago
81.3B parameters, 3B active (10 of 512 experts per token), text in, text out, 256K tokens of context. 4 providers serve it from $0.072 per million input tokens.
Verified 2026-10-25 Rank #50 in LLMs · score 107.9 method v1 Sources 5 Downloads 30d 153.5k · #269 of 619 tracked
Changed this week: Cheapest input price fell 20%: $0.090 → $0.072 per 1M (OpenRouter).2026-10-24 · 1 new quantized build on the Hub (78 in all).2026-10-24
Picked by rule from 18 tracked pieces: the lab's own post, the most shared, a field report, the latest.
OpenAI released gpt-oss-120b and gpt-oss-20b, their first open-weight models since GPT-2 in 2019.
The 2 picks above, plus these 16. 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.
3 providers with a live price, from $0.072 to $0.15 per million input tokens; the median is $0.10. One dot per provider, cheapest in teal.
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 4 API providers; heavily run locally (78 quantized builds); a big base for builders (57 fine-tunes, adapters and merges).
Top Spaces running it: hadadxyz/ai ♥ 819 · LPX55/Qwen-Image-Edit-2511-Turbo-Lightning ♥ 121 · jairwaal/image ♥ 16 · Gameyguy/Qwen-Image-Edit-2511-Turbo-Lightning ♥ 12
Weights only, an estimate: leave room for the KV cache. Method
About 49 GB at INT4, 98 GB at FP8, 195 GB at BF16: weights only, so leave room for the KV cache. Every expert must be resident, so total parameters count.
Estimate params × bytes per param × 1.2 · how we estimate
Claim-matched tests, graded by code. Receipts, not a score.
{
"providers": [
{
"provider": "parasail",
"endpoint_model": "parasail-qwen-3-next-80b-instruct",
"source_url": "https://api.parasail.io/v1/chat/completions",
"ttft_ms_median": 562,
"ttft_any_ms_median": 562,
"tokens_per_s_median": 112.1,
"runs": [
{
"ttft_s": 0.548,
"ttft_any_s": 0.548,
"tokens_per_s": 112.1,
"latency_s": 1.62,
"tokens_out": 182,
"reasoning_chars": 0,
"finish": "stop"
},
{
"ttft_s": 0.562,
"ttft_any_s": 0.562,
"tokens_per_s": 131.5,
"latency_s": 1.41,
"tokens_out": 185,
"reasoning_chars": 0,
"finish": "stop"
},
{
"ttft_s": 1.375,
"ttft_any_s": 1.375,
"tokens_per_s": 98.2,
"latency_s": 2.21,
"tokens_out": 217,
"reasoning_chars": 0,
"finish": "stop"
}
]
}
],
"note": "Single client, wall clock; tok/s counts every output token including hidden reasoning."
} {
"calls": [
{
"name": "get_weather",
"args": {
"city": "Lisbon",
"date": "2026-10-14"
}
},
{
"name": "check_calendar",
"args": {
"date": "2026-10-14"
}
},
{
"name": "book_table",
"args": {
"restaurant": "Cervejaria Ramiro",
"date": "2026-10-14",
"time": "19:30",
"party_size": 2
}
}
],
"checks": {
"get_weather": true,
"check_calendar": true,
"book_table": true
},
"final": "Your table at Cervejaria Ramiro has been confirmed for Wednesday, 2026-10-14 at 19:30 for 2 people. Confirmation code: TBL-4471. The weather in Lisbon is clear with a very low chance of rain, so your evening plans are set!",
"fixture": "receipts.errand.v1"
} {
"fields_matched": 79,
"fields_total": 80,
"cost_per_1000_usd": 0.0421,
"parsed": true,
"per_item": [
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": false,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
},
{
"street": true,
"city": true,
"postcode": true,
"country": true
}
],
"sample": [
{
"street": "1600 Amphitheatre Parkway",
"unit": null,
"city": "Mountain View",
"region": "CA",
"postcode": "94043",
"country": "US"
},
{
"street": "10 Downing Street",
"unit": null,
"city": "London",
"region": null,
"postcode": "SW1A 2AA",
"country": "GB"
},
{
"street": "Rua Augusta 274",
"unit": null,
"city": "Lisboa",
"region": null,
"postcode": "1100-053",
"country": "PT"
}
],
"fixture": "receipts.addresses.v1"
} Weights, every dated article, and repo commits on one rail.
Teal is the weights (2025-09-09), amber triangles are 5 dated articles (each links to it), grey ticks are repo commits.
135 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.
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.
| Pipeline tag | text-generation |
|---|---|
| Library | transformers |
| Architectures | Qwen3NextForCausalLM |
| Model type | qwen3_next |
| Dtype | bfloat16 |
| Params by dtype | BF16: 81,324,862,720 |
| Size band | XL |
| MoE | yes |
| Experts | 512 |
| Experts per token | 10 |
| Context length | 262,144 (config.json max_position_embeddings) |
| Papers | arXiv:2309.00071, arXiv:2404.06654, arXiv:2505.09388, arXiv:2501.15383 |
| License | Apache 2.0 (apache-2.0) |
| Hub snapshot | 2026-10-25 |
vLLM: supported · Qwen3NextForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · Qwen3NextForCausalLM · 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 Qwen3-Next-80B-A3B-Instruct: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 195 GB · Multi-GPU node (<640 GB) |
|---|---|
| FP8 (1 byte/param) | 98 GB · Multi-GPU node (<640 GB) |
| INT4 (0.5 bytes/param) | 49 GB · Single high-end GPU (<80 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra. MoE: all experts must be resident, so total parameters count.
quantized: 78finetune: 40adapter: 16merge: 1
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| Qwen/Qwen3-Next-80B-A3B-Instruct-FP8 | quantized | -FP8 | 81.3B | XL | apache-2.0 | Qwen | 98 | 3.3M |
| cyankiwi/Qwen3-Next-80B-A3B-Instruct-AWQ-4bit | quantized | AWQ | 81.3B | XL | apache-2.0 | cyankiwi | 68 | 1.9M |
| tiny-random/qwen3-next-moe | finetune | — | 2.84M | S | — | tiny-random | 4 | 676.4k |
| unsloth/Qwen3-Next-80B-A3B-Instruct-bnb-4bit | quantized | -bnb | — | — | apache-2.0 | unsloth | 26 | 531.8k |
| unsloth/Qwen3-Next-80B-A3B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | unsloth | 201 | 510.2k |
| lmstudio-community/Qwen3-Next-80B-A3B-Instruct-MLX-4bit | quantized | MLX | 79.7B | XL | apache-2.0 | lmstudio-community | 7 | 432.9k |
| lmstudio-community/Qwen3-Next-80B-A3B-Instruct-MLX-8bit | quantized | MLX | 79.7B | XL | apache-2.0 | lmstudio-community | 3 | 406.3k |
| lmstudio-community/Qwen3-Next-80B-A3B-Instruct-MLX-6bit | quantized | MLX | 79.7B | XL | apache-2.0 | lmstudio-community | 0 | 395k |
| lmstudio-community/Qwen3-Next-80B-A3B-Instruct-MLX-5bit | quantized | MLX | 79.7B | XL | apache-2.0 | lmstudio-community | 0 | 392.9k |
| nvidia/Qwen3-Next-80B-A3B-Instruct-NVFP4 | quantized | NVFP4 | — | — | apache-2.0 | nvidia | 46 | 353.1k |
| TheClusterDev/Qwen3-Next-80B-A3B-Instruct-FP8-Dynamic | quantized | -FP8 | 79.8B | XL | — | TheClusterDev | 4 | 242.8k |
| cyankiwi/Qwen3-Next-80B-A3B-Instruct-AWQ-8bit | quantized | AWQ | 81.3B | XL | apache-2.0 | cyankiwi | 4 | 213.4k |
| DevQuasar/Qwen.Qwen3-Next-80B-A3B-Instruct-FP8-Dynamic | quantized | -FP8 | 79.8B | XL | — | DevQuasar | 4 | 178.6k |
| bartowski/Qwen_Qwen3-Next-80B-A3B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | bartowski | 22 | 144.6k |
| lefromage/Qwen3-Next-80B-A3B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | lefromage | 47 | 135.5k |
| mlx-community/Qwen3-Next-80B-A3B-Instruct-4bit | quantized | -4bit | 79.7B | XL | apache-2.0 | mlx-community | 25 | 108.9k |
| mlx-community/Qwen3-Next-80B-A3B-Instruct-8bit | quantized | -8bit | 79.7B | XL | apache-2.0 | mlx-community | 10 | 103.6k |
| lmstudio-community/Qwen3-Next-80B-A3B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | lmstudio-community | 1 | 94.7k |
| RedHatAI/Qwen3-Next-80B-A3B-Instruct-NVFP4 | quantized | NVFP4 | — | — | apache-2.0 | RedHatAI | 5 | 82.6k |
| RESMP-DEV/Qwen3-Next-80B-A3B-Instruct-NVFP4 | quantized | NVFP4 | — | — | apache-2.0 | RESMP-DEV | 8 | 42.9k |
| Intel/Qwen3-Next-80B-A3B-Instruct-int4-mixed-AutoRound | quantized | int4 | 11.5B | L | apache-2.0 | Intel | 24 | 38.1k |
| RedHatAI/Qwen3-Next-80B-A3B-Instruct-quantized.w4a16 | quantized | -quant | 12.2B | L | apache-2.0 | RedHatAI | 4 | 35.9k |
| yujiepan/qwen3-next-moe-tiny-random | finetune | — | 2.84M | S | — | yujiepan | 0 | 33.3k |
| RedHatAI/Qwen3-Next-80B-A3B-Instruct-FP8 | quantized | -FP8 | 81.3B | XL | apache-2.0 | RedHatAI | 0 | 25.9k |
| TheClusterDev/Qwen3-Next-80B-A3B-Instruct-FP8 | quantized | -FP8 | 79.8B | XL | — | TheClusterDev | 1 | 23k |
| unsloth/Qwen3-Next-80B-A3B-Instruct | finetune | — | 81.3B | XL | apache-2.0 | unsloth | 88 | 10.7k |
| dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16 | quantized | GPTQ | 14.4B | L | — | dazipe | 5 | 12.8k |
| mradermacher/Qwen3-Next-80B-A3B-Instruct-i1-GGUF | quantized | GGUF | — | — | apache-2.0 | mradermacher | 2 | 12.6k |
| Intel/Qwen3-Next-80B-A3B-Instruct-int4-AutoRound | quantized | int4 | 11.2B | L | apache-2.0 | Intel | 10 | 7.6k |
| lovedheart/Qwen3-Next-REAP-40B-A3B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | lovedheart | 1 | 8.1k |
| ilintar/Qwen3-Next-80B-A3B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | ilintar | 11 | 7.5k |
| jart25/Qwen3-Next-80B-A3B-Instruct-Int4-GPTQ | quantized | Int4 | — | — | — | jart25 | 3 | 6.1k |
| mradermacher/Qwen3-Next-80B-A3B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | mradermacher | 2 | 5.5k |
| noctrex/Qwen3-Next-80B-A3B-Instruct-MXFP4_MOE-GGUF | quantized | GGUF | — | — | — | noctrex | 7 | 4.6k |
| lovedheart/Qwen3-Next-REAP-25B-A3B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | lovedheart | 2 | 4.6k |
| lovedheart/Qwen3-Next-REAP-30B-A3B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | lovedheart | 4 | 3.8k |
| John1604/Qwen3-Next-80B-A3B-Instruct-gguf | quantized | gguf | — | — | apache-2.0 | John1604 | 0 | 3k |
| mlx-community/Qwen3-Next-80B-A3B-Instruct-6bit | quantized | — | 79.7B | XL | apache-2.0 | mlx-community | 4 | 2.2k |
| RedHatAI/Qwen3-Next-80B-A3B-Instruct-FP8-dynamic | quantized | -FP8 | 79.8B | XL | apache-2.0 | RedHatAI | 1 | 2.4k |
| LibraxisAI/Qwen3-Next-80B-A3B-Instruct-MLX-MXFP4 | quantized | MLX | 79.7B | XL | apache-2.0 | LibraxisAI | 2 | 2.2k |
| Date | Commit | SHA |
|---|---|---|
| 2025-09-17 | Update tokenizer_config.json | 9c7f2fb |
| 2025-09-15 | Update README.md | 3eb90af |
| 2025-09-15 | Update README.md | 92d3dc0 |
| 2025-09-12 | Update README.md | 609718e |
| 2025-09-11 | Update README.md | cfdff41 |
| 2025-09-11 | Upload tokenizer.json | c7646fb |
| 2025-09-11 | Delete tokenizer.json | 7cec4bd |
| 2025-09-11 | Create LICENSE | e62a222 |
| 2025-09-11 | Update generation_config.json | 9aeffe7 |
| 2025-09-11 | Update config.json | 1f3f9fc |
| 2025-09-11 | Update README.md | f7597ba |
| 2025-09-09 | Add files using upload-large-folder tool | 21dee27 |
| 2025-09-09 | initial commit | 219812d |
Hosted endpoints observed for Qwen3-Next-80B-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.
| Provider | Model ref | Context | Quant | $/M in | $/M out | tok/s | TTFT | Tools | Status | Observed |
|---|---|---|---|---|---|---|---|---|---|---|
| DeepInfra hf, deepinfra, openrouter | Qwen/Qwen3-Next-80B-A3B-Instruct | 256K | fp8 | $0.07 | $0.88 | 120.5 | 545 ms | yes | live | 2026-10-25 |
| Parasail parasail, openrouter | parasail-qwen-3-next-80b-instruct | 256K | fp8 | $0.10 | $1.10 | — | — | yes | serverless | 2026-10-24 |
| Google openrouter | qwen/qwen3-next-80b-a3b-instruct | 256K | unknown | $0.15 | $1.20 | — | — | yes | live | 2026-10-24 |
| Featherless hf | Qwen/Qwen3-Next-80B-A3B-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 21:32 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
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.