Qwen3 4B 2507 (Reasoning)
Qwen3 4B 2507 (Reasoning)Ranking?
Alternatives?
CheaperClaude Haiku 5.5max×4.9-60%$0.21Remove sampling parameters+5
BetterGPT-6 Lunalow×2.4-60%$0.005Function calling only with none · Chat Completions+3IOpenK2 Horizon 7Bthinking×2.3+5%$0.30 / $1.20Memory (4-bit): 2 GB → 4.5 GB
CheaperMiniMax M3.1 Flashthinking×4.9-48%$0.22Complete model response must be appended
CheaperGLM 5.3 FlashmaxOpen×4.8-53%$0.28Memory (4-bit): 2 GB → 193 GB+5Claude Haiku 5.5max×4.9-60%$0.21Remove sampling parameters+5MiniMax M3.1 Flashthinking×4.9-48%$0.22Complete model response must be appendedGLM 5.3 FlashmaxOpen×4.8-53%$0.28Memory (4-bit): 2 GB → 193 GB+5
Ling 3.1 FlashOpen×4.7-10%$1.00Memory (4-bit): 2 GB → 343 GBQwen3.8 FlashthinkingOpen×4.5-54%$0.37Memory (4-bit): 2 GB → 108 GB+6GPT-6 Lunalow×2.4-60%$0.005Function calling only with none · Chat Completions+3GPT-5.6 Lunalow×2.4-10%$0.01
Mistral Small 4thinkingOpen+28%-48%$0.015Memory (4-bit): 2 GB → 72 GB
MiMo v2.5thinkingOpen×2.5-65%$0.019Memory (4-bit): 2 GB → 186 GB
Granite 4.2 8BthinkingOpen+26%-79%$0.024Context: 262K → 128K+1
Gemma 4 12BthinkingOpen+61%-70%$0.10 / $0.30Context: 262K → 256K+1Qwen 3.5 9Bnon-thinkingOpen+51%-82%$0.08 / $0.13Memory (4-bit): 2 GB → 4.8 GBQwen3.5 4BthinkingOpen+49%-88%$0.03 / $0.15Memory (4-bit): 2 GB → 2.3 GBQwen3.8 Max (0902)thinking×5.2—$5.41Qwen3.8 2.4T A95BthinkingOpen×4.5—$2.16Memory (4-bit): 2 GB → 1,468 GB+5Qwen3.8 FlashthinkingOpen×4.5—$0.37Memory (4-bit): 2 GB → 108 GB+6Qwen3.7 Maxthinking×3.3—$1.15Qwen3.6 Max Previewthinking×3.2—$1.30 / $7.80Include the word JSON · QwenCloudQwen3.8 27BmediumOpen×3.1—$1.13Memory (4-bit): 2 GB → 17 GB+5Qwen3.6 Plusthinking×3.1—$0.50 / $3Qwen3.7 Plusthinking×2.9—$0.22All Alibaba (Qwen) models →Run it yourself
estimateHistory
3 versionsBenchmarks
all categoriesFAQ
How much does Qwen3 4B 2507 (Reasoning) cost?›
$0.40 per 1M input tokens and $0.80 per 1M output tokens (cheapest host: Featherless).
Is Qwen3 4B 2507 (Reasoning) free to use?›
There is no free API; the cheapest paid price is $0.40 per 1M input tokens. The weights are open, so you can run it yourself.
How good is Qwen3 4B 2507 (Reasoning), and what is it best at?›
#245 of 334 in Chat & Knowledge, #195 of 239 in Coding, #170 of 242 in Agents & Tool Use, #117 of 245 in Writing & Language.
Is Qwen3 4B 2507 (Reasoning) better than Qwen3 4B?›
On AA Intelligence Index 4.3.2, Qwen3 4B 2507 (Reasoning) scores 8.8 vs 7.2 — higher.
How can I use Qwen3 4B 2507 (Reasoning)?›
Through the API of Featherless.
What is the context window of Qwen3 4B 2507 (Reasoning)?›
262K tokens.
When was Qwen3 4B 2507 (Reasoning) released? Is there a newer version?›
Released 2025-08-06. It has been superseded by Qwen3.5 4B.
Is Qwen3 4B 2507 (Reasoning) open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run Qwen3 4B 2507 (Reasoning) locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Qwen3 4B 2507 (Reasoning) read images?›
No, it accepts text only.
What is Qwen3 4B 2507 (Reasoning)?›
Qwen3 4B 2507 (Reasoning) is an open-weights language model from Alibaba (Qwen), released Aug 2025, the successor to Qwen3 4B.
What are the best alternatives to Qwen3 4B 2507 (Reasoning)?›
Claude Haiku 5.5 (cheaper), GPT-6 Luna (better), K2 Horizon 7B (open weights).