Mistral Small 4
Mistral Small 4Alternatives?
CheaperLing 3.0 Flash VLthinkingOpenwins 99%-66%$0.06 / $0.18Memory (4-bit): 72 GB → 75 GB+1
BetterGPT-6 Lunalowwins 95%-70%$0.005Function calling only with none · Chat Completions+3
OpenQwen3.8 Flashthinkingwins 100%+2,373%$0.37Memory (4-bit): 72 GB → 108 GB+6CheaperLing 3.0 Flash FinthinkingOpenwins 98%-66%$0.06 / $0.18Memory (4-bit): 72 GB → 76 GB+2CheaperLing 3.0 FlashthinkingOpenwins 96%-66%$0.06 / $0.18Memory (4-bit): 72 GB → 76 GB+2Ling 3.0 Flash VLthinkingOpenwins 99%-66%$0.06 / $0.18Memory (4-bit): 72 GB → 75 GB+1Ling 3.0 Flash FinthinkingOpenwins 98%-66%$0.06 / $0.18Memory (4-bit): 72 GB → 76 GB+2GPT-6 Lunanon-thinkingwins 98%-25%$0.011Function calling only with none · Chat Completions+3Ling 3.0 FlashthinkingOpenwins 96%-66%$0.06 / $0.18Memory (4-bit): 72 GB → 76 GB+2GPT-5.6 Lunalowwins 95%-35%$0.01GPT-6 Lunalowwins 95%-70%$0.005Function calling only with none · Chat Completions+3GPT-5.6 Lunalowwins 95%-35%$0.01
MiMo v2.5thinkingOpenwins 95%+27%$0.019Memory (4-bit): 72 GB → 186 GB+1
Granite 4.2 8BthinkingOpenwins 60%+57%$0.024Context: 262K → 128K+2
Claude Haiku 5.5lowwins 99%+63%$0.024Remove sampling parameters+5Qwen3.8 FlashthinkingOpenwins 100%+2,373%$0.37Memory (4-bit): 72 GB → 108 GB+6Qwen3.8 27BxhighOpenwins 100%+6,596%$1.01max_tokens range · QwenCloud+4IK2 Horizon MoVA 36B A4BthinkingOpenwins 100%+567%$1 / $4No image inputIK2 Horizon 7BthinkingOpenwins 99%+100%$0.30 / $1.20No image input
Nemotron 3.5 Super VLOpenwins 99%+1,151%$0.19Memory (4-bit): 72 GB → 75 GB+1Mistral Large 4thinkingwins 100%+7,437%$1.13Infinite whitespace stream+4Mistral Medium 3.5highOpenwins 84%+3,237%$0.50Memory (4-bit): 72 GB → 77 GBMistral Large 3non-thinkingOpensame+107%$0.031Memory (4-bit): not verifiedMinistral 3 14B 2512non-thinkingOpenwins 35%+29%$0.019Ministral 3 8B 2512Openwins 28%-26%$0.011Ministral 3 3B 2512Openwins 20%-48%$0.008Context: 262K → 131KMistral Smalln/a—$1 / $3Context: 262K → 32K+1Ministral 14B 3.0n/a—$0.20 / $0.20Context: 262K → 128K+1All Mistral models →Run it yourself
estimateBenchmarks
in Agents & Tool UseFAQ
How much does Mistral Small 4 cost?›
$0.15 per 1M input tokens and $0.60 per 1M output tokens; cached input $0.015 per 1M. A typical benchmark task costs about $0.015 at default effort.
Is Mistral Small 4 free to use?›
There is no free API; the cheapest paid price is $0.15 per 1M input tokens. The weights are open, so you can run it yourself.
How good is Mistral Small 4, and what is it best at?›
#193 of 334 in Chat & Knowledge, #75 of 132 in Coding, #112 of 133 in Agents & Tool Use, #73 of 95 in Writing & Language, #101 of 124 in Vision & Documents.
Is Mistral Small 4 better than Trinity Large Thinking?›
On AA-Briefcase 1.1 · Elo, Mistral Small 4 scores 241.1 vs 255.5 — about the same. It is 87% cheaper per task.
How can I use Mistral Small 4?›
Through the API of Mistral.
What is the context window of Mistral Small 4?›
262K tokens, with up to 210K output tokens.
When was Mistral Small 4 released? Is there a newer version?›
Released 2026-03-16. It is the current version in its line.
Is Mistral Small 4 open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run Mistral Small 4 locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Mistral Small 4 read images?›
It accepts text and images, PDF files as input.
What is Mistral Small 4?›
Mistral Small 4 is an open-weights language model from Mistral, released Mar 2026, the successor to Saba.
What are the best alternatives to Mistral Small 4?›
Ling 3.0 Flash VL (cheaper), GPT-6 Luna (better), Qwen3.8 Flash (open weights).