Skip to content
AI Atlas
Familyroot · Qwen

Qwen3

All 61 models in the terminal →

first seen 11 Sept 2026

family_01M2A55V0V38EPYN9HB7MPQHZF

Models
61
First release
27 Apr 2025
Latest release
14 May 2026
Parameters
595.8M – 480B
Artifacts
8
Providers
1

Releases

Family releases by statusActiveAnnouncedDeprecatedMay 25Jun 25Jul 25Aug 25Sept 25Oct 25Nov 25Dec 25Jan 26Feb 26Mar 26Apr 26May 26Qwen3 32B — 27 Apr 2025Qwen3 8B — 27 Apr 2025Qwen3-4B — 27 Apr 2025Qwen3 14B — 28 Apr 2025Qwen3 30B A3B — 28 Apr 2025Qwen3 235B A22B — 29 Apr 2025DeepSeek-R1-0528-Qwen3-8B — 29 May 2025Qwen3-Embedding-0.6B — 3 Jun 2025Qwen3 Embedding Series — 5 Jun 2025Qwen3 235B A22B Instruct 2507 — 21 Jul 2025qwen3-coder-480b-a35b-instruct — 22 Jul 2025Qwen3 Coder 480B A35B — 23 Jul 2025Qwen3 235B A22B Thinking 2507 — 25 Jul 2025Qwen3 30B A3B Instruct 2507 — 29 Jul 2025Qwen3 Coder 30B A3B Instruct — 31 Jul 2025Qwen3 30B A3B Thinking 2507 — 28 Aug 2025Qwen3 Next 80B A3B Instruct — 11 Sept 2025Qwen3 Next 80B A3B Thinking — 11 Sept 2025Qwen3 Coder Flash — 17 Sept 2025Qwen3 Coder Plus — 23 Sept 2025Qwen3 Max — 23 Sept 2025Qwen3 VL 235B A22B Instruct — 23 Sept 2025Qwen3 VL 235B A22B Thinking — 23 Sept 2025Qwen3Guard — 23 Sept 2025Qwen3 VL 30B A3B Instruct — 6 Oct 2025Qwen3 VL 30B A3B Thinking — 6 Oct 2025Qwen3 VL 8B Instruct — 11 Oct 2025Qwen3-VL-4B-Instruct — 11 Oct 2025Qwen3 VL 8B Thinking — 14 Oct 2025cerebras/Qwen3-Coder-REAP-25B-A3B — 20 Oct 2025Qwen3 VL 32B Instruct — 23 Oct 2025Qwen3 Coder Next — 4 Feb 2026amazon/Qwen3-Coder-30B-A3B-Instruct-P-EAGLE-long-context — 14 May 2026

33 dated releases · dot size = best benchmark rank of the member · 28 members without a sourced release date not shown

Members 61

Family members
ModelParamsContextOpennessLicenceReleasedStatusBest rankCompare
amazon/Qwen3-Coder-30B-A3B-Instruct-P-EAGLE-long-contextAmazon Web Services972.9M3B active—Open weightsApache-2.014 May 2026Active—
Qwen3 Coder NextQwen—262.1KOpen weights—4 Feb 2026Active#98 SciCode
Qwen3 VL 32B InstructQwen—131.1KOpen weights—23 Oct 2025Deprecated#93 MMMU-Pro
cerebras/Qwen3-Coder-REAP-25B-A3BCerebras Systems24.9B3B active—Open weightsApache-2.020 Oct 2025Active—
Qwen3 VL 8B ThinkingQwen—131.1KOpen weights—14 Oct 2025Active—
Qwen3 VL 8B InstructQwen8.77B262.1KOpen weightsApache-2.011 Oct 2025Deprecated#115 MMMU-Pro
Qwen3-VL-4B-InstructQwen4.44B256KOpen weightsApache-2.011 Oct 2025Deprecated#125 MMMU-Pro
Qwen3 VL 30B A3B InstructQwen—262.1KOpen weights—6 Oct 2025Deprecated#100 MMMU-Pro
Qwen3 VL 30B A3B ThinkingQwen—262.1KOpen weights—6 Oct 2025Active—
Qwen3 Coder PlusQwen—1M——23 Sept 2025Active—
Qwen3 MaxQwen—262.1KClosed—23 Sept 2025Deprecated#57 IFBench
Qwen3 VL 235B A22B InstructQwen—262.1KOpen weights—23 Sept 2025Deprecated#79 MMMU-Pro
Qwen3 VL 235B A22B ThinkingQwen—131.1KOpen weights—23 Sept 2025Active—
Qwen3GuardQwen Team——Open weights—23 Sept 2025Announced—
Qwen3 Coder FlashQwen—1M——17 Sept 2025Active—
Qwen3 Next 80B A3B InstructQwen—262.1KOpen weights—11 Sept 2025Active#101 IFBench
Qwen3 Next 80B A3B ThinkingQwen—262.1KOpen weights—11 Sept 2025Active—
Qwen3 30B A3B Thinking 2507Qwen—81.9KOpen weights—28 Aug 2025Active—
Qwen3 Coder 30B A3B InstructQwen—262.1KOpen weights—31 Jul 2025Deprecated#143 Terminal-Bench
Qwen3 30B A3B Instruct 2507Qwen—262.1KOpen weights—29 Jul 2025Active—
Qwen3 235B A22B Thinking 2507Qwen—131.1KOpen weights—25 Jul 2025Active—
Qwen3 Coder 480B A35BQwen—262.1KOpen weights—23 Jul 2025Active—
qwen3-coder-480b-a35b-instructAlibaba Group480B35B active256KOpen weights—22 Jul 2025Announced#25 SWE-bench Verified
Qwen3 235B A22B Instruct 2507Qwen—262.1KOpen weights—21 Jul 2025Deprecated#76 SciCode
Qwen3 Embedding SeriesQwen Team—32.8KOpen weightsApache-2.05 Jun 2025Active—
Qwen3-Embedding-0.6BQwen595.8M—Open weightsApache-2.03 Jun 2025Active—
DeepSeek-R1-0528-Qwen3-8BDeepSeek8.19B32.8KOpen weightsMIT29 May 2025Deprecated#253 Terminal-Bench
Qwen3 235B A22BQwen235B22B active131.1KOpen weightsApache-2.029 Apr 2025Active—
Qwen3 14BQwen—131.1KOpen weights—28 Apr 2025Deprecated#110 SciCode
Qwen3 30B A3BQwen—131.1KOpen weights—28 Apr 2025Active—
Qwen3 32BQwen32.8B131.1KOpen weightsApache-2.027 Apr 2025Deprecated#37 Aider polyglot
Qwen3 8BQwen8.19B131.1KOpen weightsApache-2.027 Apr 2025Deprecated#113 SciCode
Qwen3-4BQwen4.02B32KOpen weightsApache-2.027 Apr 2025Deprecated#257 τ²-bench
Qwen3—————Active—
Qwen3 235B A22B diff, no think, Alibaba API—————Active#20 Aider polyglot
qwen3-14b-instructAlibaba Group—32.8KOpen weights——Deprecated#184 τ²-bench
qwen3-235b-a22b-instructAlibaba Group—32.8KOpen weights——Deprecated#183 Humanity's Last Exam
qwen3-30b-a3b-2507Alibaba Group—262.1KOpen weights——Deprecated#103 SciCode
qwen3-30b-a3b-instructAlibaba Group—32.8KOpen weights——Deprecated#184 Terminal-Bench
qwen3-32b-instructAlibaba Group—32.8KOpen weights——Deprecated#266 IFBench
qwen3-4b-2507-instructAlibaba Group—262.1KOpen weights——Deprecated#146 IFBench
Qwen3-4B-BaseQwen——Open weights——Active—
qwen3-4b-instructAlibaba Group—32KOpen weights——Deprecated#355 AA Intelligence Index
Qwen3-4B-Instruct-2507Qwen——Open weights——Active—
Qwen3-8B-BaseQwen——Open weights——Active—
qwen3-8b-instructAlibaba Group—32.8KOpen weights——Deprecated#223 τ²-bench
Qwen3-Embedding-4BQwen Team—————Active—
Qwen3-Embedding-8BQwen Team—————Active—
qwen3-max-previewAlibaba Group—262.1KClosed——Deprecated#122 Terminal-Bench
qwen3-max-thinking-previewAlibaba Group—262.1KClosed——Deprecated#87 τ²-bench
qwen3-omni-30b-a3b-instructAlibaba Group—65.5KOpen weights——Active#107 MMMU-Pro
Qwen3-Reranker-0.6BQwen Team—————Active—
Qwen3-Reranker-4BQwen Team—————Active—
Qwen3-Reranker-8BQwen Team—————Active—
Qwen3-VL-2B-InstructQwen——Open weights——Active—
Qwen3Guard-Gen-0.6BQwen Team—————Active—
Qwen3Guard-Gen-4BQwen Team—————Active—
Qwen3Guard-Gen-8BQwen Team—————Active—
Qwen3Guard-Stream-0.6BQwen Team—————Active—
Qwen3Guard-Stream-4BQwen Team—————Active—
Qwen3Guard-Stream-8BQwen Team—————Active—

Lineage among members 7 relations

Open in Graph →
Lineage of the Qwen3 familycerebras/Qwen3-Coder-REAP-25B-A3B fine-tuned from Qwen3 Coder 30B A3B InstructQwen3 8B fine-tuned from Qwen3-8B-BaseQwen3-4B fine-tuned from Qwen3-4B-BaseQwen3 superseded by Qwen3GuardQwen3 Embedding Series derived from Qwen3Qwen3Guard derived from Qwen3Qwen3 superseded by Qwen3 Embedding SeriesQwen3 8B8.19BQwen3 8BQwen3-4B4.02BQwen3-4BQwen3 Embedding SeriesQwen3 Embedding SeriesQwen3 Coder 30B A3B Instr…Qwen3 Coder 30B A3B InstructQwen3GuardQwen3Guardcerebras/Qwen3-Coder-REAP…24.9Bcerebras/Qwen3-Coder-REAP-25B-A3BQwen3Qwen3Qwen3-4B-BaseQwen3-4B-BaseQwen3-8B-BaseQwen3-8B-Base

Benchmark progress

IFBenchbest #57

Best rank of the family on IFBench#300#200#100#0Jul 25Oct 25Jan 26

SciCodebest #76

Best rank of the family on SciCode#120#100#80#60#40#20#0Jul 25Oct 25Jan 26

MMMU-Probest #79

Best rank of the family on MMMU-Pro#100#80#60#40#20#0Sept 25Oct 25Oct 25Oct 25

GPQA Diamondbest #82

Best rank of the family on GPQA Diamond#300#200#100#0Jul 25Oct 25Jan 26

Humanity's Last Exambest #87

Best rank of the family on Humanity's Last Exam#300#200#100#0Jul 25Oct 25Jan 26

τ²-benchbest #87

Best rank of the family on τ²-bench#250#200#150#100#50#0Jul 25Oct 25Jan 26

Best rank (primary comparability group, current leaderboard) reached by any member released up to each date — lower is better, axis inverted. Ranks are today's ranks, not the ranks at release time.

Aggregates are computed from the members' sourced claims; members without a release date or parameter count are left out of the range and timeline rather than guessed. Methodology →