Qwen 2.5
Qwen 2.5Compare inon
Benchmarks of Qwen 2.515
BenchmarkByScoreRank
No benchmark matches
—per task?
API price not publicprice
—output speed?
—context
Ranking?
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-6 Lunamaxwins 58%—$0.068Function calling only with none · Chat Completions+3
OpenKimi K3maxwins 65%—$2.00Sampling parameters are fixed+3BetterGPT-5.6 Lunaxhighwins 58%—$0.085BetterGPT-5.6 Terraxhighwins 61%—$0.63Prompts with >272K input tokensBetterGPT-6.1 Solmaxwins 65%—$0.72Remove unsupported parameters+5ModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-6 Lunamaxwins 58%—$0.068Function calling only with none · Chat Completions+3GPT-5.6 Lunaxhighwins 58%—$0.085GPT-5.6 Terraxhighwins 61%—$0.63Prompts with >272K input tokensGPT-6.1 Solmaxwins 65%—$0.72Remove unsupported parameters+5
Muse Spark 1.2xhighwins 63%—$0.97Only auto tool_choice is supported+1ModelScoreCostPer taskWhat changes
Kimi K3maxOpenwins 65%—$2.00Sampling parameters are fixed+3
GLM 5.2maxOpenwins 64%—$1.47GLM 5.3maxOpenwins 64%—$2.01Reasoning cannot be disabled+3GLM 5unstatedOpenwins 63%—$1 / $3.20GLM 5.1unstatedOpenwins 63%—$1.40 / $4.40ModelScoreCostPer taskWhat changes
Qwen 3.7 Max Previewwins 62%—$2.50 / $7.50Qwen3.6 Max Previewunstatedwins 62%—$1.30 / $7.80Include the word JSON · QwenCloudQwen3.7 Plusunstatedwins 61%—$0.32 / $1.28Qwen3.6 Plusunstatedwins 58%—$0.50 / $3Qwen3.5 397B A17BunstatedOpenwins 58%—$0.60 / $3.60Qwen3 Max (Preview)unstatedwins 56%—$0.87 / $3.48Qwen 3Openwins 56%——Qwen3 235B A22B Thinking 2507unstatedOpenwins 52%—$0.23 / $2.30No strict JSONAll Alibaba (Qwen) models →Run it yourself
Fits onhardware needs not verified yet
Licencenot verified
Benchmarks
all categoriesChat & Knowledge
Arena Text · overall1,374#111
Arena Text · Chinese1,381.8#115
Arena Text · English1,375.6#116
Arena Text · French1,396.2#92
Arena Text · German1,349.6#100
Arena Text · instruction following1,334.7#111
Arena Text · Japanese1,299.7#88
Arena Text · Korean1,303.6#96
Arena Text · non-English1,352.2#106
Arena Text · Polish1,361.1#89
Arena Text · Russian1,352.9#107
Arena Text · Spanish1,376.9#98
MMLU-Pro · official boardlab76.1
Arena Text · overall · full cohort1,366.7#107
Writing & Language
Arena Text · creative writing1,353#100
Arena Text · creative writing · full cohort1,338.5#100
FAQ
How much does Qwen 2.5 cost?›
There is no per-unit API price: API price not public.
Is Qwen 2.5 free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Qwen 2.5, and what is it best at?›
#111 of 196 in Chat & Knowledge, #100 of 195 in Writing & Language.
Is Qwen 2.5 better than Hy3 preview?›
On Arena Text · creative writing, Qwen 2.5 scores 1,353 vs 1,354 — about the same.
Is Qwen 2.5 open source?›
Yes, the weights are open.
Can I run Qwen 2.5 locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Qwen 2.5 read images?›
No, it accepts text only.
What is Qwen 2.5?›
Qwen 2.5 is an open-weights language model from Alibaba (Qwen).
What are the best alternatives to Qwen 2.5?›
GPT-6 Luna (better), Kimi K3 (open weights).