Skip to content
AI Atlas
Familyroot · Qwen

Qwen2.5

All 47 models in the terminal →

first seen 11 Sept 2026

family_01M2A55V383Q144TZ1DJWP30GT

Models
47
First release
16 Sept 2024
Latest release
27 Mar 2025
Parameters
494M – 72.7B
Artifacts
2
Providers
2

Releases

Family releases by statusActiveDeprecatedOtherOct 24Nov 24Dec 24Jan 25Feb 25Mar 25Qwen2.5 7B Instruct — 16 Sept 2024Qwen2.5-0.5B-Instruct — 16 Sept 2024Qwen2.5-1.5B-Instruct — 17 Sept 2024Qwen2.5-3B-Instruct — 17 Sept 2024Qwen2.5 — 19 Sept 2024Qwen2.5 72B Instruct — 19 Sept 2024Qwen2.5-Math — 19 Sept 2024Qwen2.5 Coder 32B Instruct — 11 Nov 2024Qwen2.5-Coder — 12 Nov 2024Qwen2.5 VL — 26 Jan 2025Qwen2.5-VL-7B-Instruct — 26 Jan 2025Qwen2.5-1M — 27 Jan 2025qwen-2-5-max — 28 Jan 2025Qwen2.5 VL 72B Instruct — 1 Feb 2025Qwen2.5-VL-32B-Instruct — 24 Mar 2025Qwen2.5-Omni — 27 Mar 2025

16 dated releases · dot size = best benchmark rank of the member · 31 members without a sourced release date not shown

Members 47

Family members
ModelParamsContextOpennessLicenceReleasedStatusBest rankCompare
Qwen2.5-OmniQwen Team7B—Open weights—27 Mar 2025Active—
Qwen2.5-VL-32B-InstructQwen Team32B—Open weightsApache-2.024 Mar 2025Active—
Qwen2.5 VL 72B InstructQwen—128KOpen weights—1 Feb 2025Active—
qwen-2-5-maxAlibaba Group—32KClosed—28 Jan 2025Active#286 AA Intelligence Index
Qwen2.5-1MQwen Team—1MOpen weights—27 Jan 2025Active—
Qwen2.5 VLQwen Team——Open weights—26 Jan 2025Active—
Qwen2.5-VL-7B-InstructQwen8.29B—Open weightsApache-2.026 Jan 2025Active—
Qwen2.5-CoderQwen Team32.5B31B active128KRestricted weightsResearch-Only12 Nov 2024open-source—
Qwen2.5 Coder 32B InstructQwen—32.8KOpen weights—11 Nov 2024Deprecated#11 Aider polyglot — well-
Qwen2.5Qwen Team72.7B70B active131.1KRestricted weightsResearch-Only19 Sept 2024Active—
Qwen2.5 72B InstructQwen—32.8KOpen weights—19 Sept 2024Deprecated#174 τ²-bench
Qwen2.5-MathQwen Team—4.1KOpen weights—19 Sept 2024Active—
Qwen2.5-1.5B-InstructQwen1.54B—Open weightsApache-2.017 Sept 2024Active—
Qwen2.5-3B-InstructQwen3.09B—Open weightsOther17 Sept 2024Active—
Qwen2.5 7B InstructQwen7.62B32.8KOpen weightsApache-2.016 Sept 2024Active—
Qwen2.5-0.5B-InstructQwen494M—Open weightsApache-2.016 Sept 2024Active—
Qwen2.5 (7B + 72B)Qwen—————Active—
Qwen2.5 7B Instruct Turbo—————Active—
Qwen2.5-0.5BQwen——Open weights——Active—
Qwen2.5-1.5BQwen——Open weights——Active—
Qwen2.5-14BQwen Team—————Active—
Qwen2.5-14B-InstructQwen Team—————Active—
Qwen2.5-14B-Instruct-1MQwen Team—————Active—
Qwen2.5-32BQwen——Open weights——Active—
qwen2.5-32b-instructAlibaba Group—128KOpen weights——Deprecated#345 AA Intelligence Index
Qwen2.5-3BQwen——Open weights——Active—
Qwen2.5-72BQwen——Open weights——Active—
Qwen2.5-7BQwen——Open weights——Active—
Qwen2.5-7B-Instruct-1MQwen Team—————Active—
Qwen2.5-Coder-0.5BQwen Team—————Active—
Qwen2.5-Coder-1.5BQwen Team—————Active—
Qwen2.5-Coder-14BQwen Team—————Active—
Qwen2.5-Coder-3BQwen Team—————Active—
Qwen2.5-Coder-7BQwen—131.1KOpen weights——Deprecated#318 Humanity's Last Exam
Qwen2.5-Coder-Base—————Active—
Qwen2.5-Coder-InstructQwen Team—————Active—
Qwen2.5-Math-1.5BQwen Team—————Active—
Qwen2.5-Math-1.5B-InstructQwen Team—————Active—
Qwen2.5-Math-72BQwen Team—————Active—
Qwen2.5-Math-72B-InstructQwen Team—————Active—
Qwen2.5-Math-7BQwen——Open weights——Active—
Qwen2.5-Math-7B-InstructQwen Team—————Active—
Qwen2.5-Math-RM-72BQwen Team—————Active—
Qwen2.5-Omni-7BQwen Team—————Active—
Qwen2.5-VL series—————Active—
Qwen2.5-VL-3BQwen Team—————Active—
QwQ-32B + Qwen 2.5 Coder Instruct—————Active#1 Aider polyglot — well-

Lineage among members 8 relations

Open in Graph →
Lineage of the Qwen2.5 familyQwen2.5 7B Instruct fine-tuned from Qwen2.5-7BQwen2.5-1.5B-Instruct fine-tuned from Qwen2.5-1.5BQwen2.5-0.5B-Instruct fine-tuned from Qwen2.5-0.5BQwen2.5-3B-Instruct fine-tuned from Qwen2.5-3BQwen2.5-Coder derived from Qwen2.5Qwen2.5-Math derived from Qwen2.5Qwen2.5-VL series superseded by Qwen2.5-VL-32B-InstructQwen2.5-Coder derived from Qwen2.5-Coder-BaseQwen2.5 7B Instruct7.62BQwen2.5 7B InstructQwen2.5-0.5B-Instruct494MQwen2.5-0.5B-InstructQwen2.5-1.5B-Instruct1.54BQwen2.5-1.5B-InstructQwen2.5-3B-Instruct3.09BQwen2.5-3B-InstructQwen2.572.7BQwen2.5Qwen2.5-MathQwen2.5-MathQwen2.5-Coder32.5BQwen2.5-CoderQwen2.5-VL-32B-Instruct32BQwen2.5-VL-32B-InstructQwen2.5-0.5BQwen2.5-0.5BQwen2.5-1.5BQwen2.5-1.5BQwen2.5-3BQwen2.5-3BQwen2.5-7BQwen2.5-7BQwen2.5-Coder-BaseQwen2.5-Coder-BaseQwen2.5-VL seriesQwen2.5-VL series

Benchmark progress

Artificial Analysis Intelligence Indexbest #286

Best rank of the family on Artificial Analysis Intelligence Index#400#300#200#100#0Oct 24Nov 24Dec 24Jan 25

GPQA Diamondbest #287

Best rank of the family on GPQA Diamond#400#300#200#100#0Oct 24Nov 24Dec 24Jan 25

Humanity's Last Exambest #400

Best rank of the family on Humanity's Last Exam#500#400#300#200#100#0Oct 24Nov 24Dec 24Jan 25

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 →